1.1 Overview of Literature Review
This paper provides a critical, integrative review of the literature covering aspects of auditory forebrain anatomy and physiology with a focus on the reciprocal corticothalamic and corticocortical connections in mammalian auditory systems. It is an integrative review in the respect that it critically reviews the existing literature that deals with reciprocal connections in the upper auditory pathway and interprets this literature in an integrative fashion in the light of a paradigm of interactive sensory processing. Special emphasis will be placed on: 1) the examination of the reentrant processing of information in the auditory cortex and its role and significance in auditory perception; 2) the role of cortical layer I in the reentrant processing of auditory information; 3) the reciprocal connections between the MGm and the auditory cortex.
1.2 List of Abbreviations Used
Oc2l occipital cortex, area 2, lateral
Oc1 occipital cortex, area 1
PRh perirhinal cortex
PAR2 parietal cortex, area 2
PAR1 parietal cortex, area 1
A anterior auditory cortical field
AAF anterior auditory cortical field
Ach acetylcholine
AEP auditory evoked potential
AI primary auditory cortical field
AII secondary auditory cortical field
AMYG amygdala
Au1 primary auditory cortex
AuD secondary auditory cortex, dorsal area
AuV secondary auditory cortex, ventral area
BIC brachium of the inferior colliculus
cd caudodorsal nucleus of the medial geniculate, dorsal division
CS conditioned stimulus
CSD current source density
d dorsal nucleus of the medial geniculate, dorsal division or dorsal
cap nucleus
DA-TR dextranamine conjugated to tetramethylrhodamine
Dc caudodorsal nucleus of the medial geniculate, dorsal division
dd (DD) deep dorsal nucleus of the medial geniculate, dorsal division
DP dorsoposterior area
DS superficial dorsal nucleus of the medial geniculate, dorsal division
DY diamidino yellow
DZ dorsal zone of the auditory cortex
Ea ectosylvian anterior auditory area
EE excitatory/excitatory binaural band
EI excitatory/inhibitory binaural band
Ep posterior ectosylvian gyrus
HRP horseradish peroxidase
I insular auditory field
IC inferior colliculus
Icc (ICC) central nucleus of the inferior colliculus
ICp pericentral nucleus of the inferior colliculus
Icx external nucleus of the inferior colliculus
IFC isofrequency contours
Ins (INS) insular auditory field
KA koniocortex
LV pars lateralis of the medial geniculate body, ventral division
M medial division of medial geniculate
MBd medial geniculate, dorsal division
MG medial geniculate
Mg medium and large cell portion of the medial division of the MGB
MGB medial geniculate body
MGcd medial geniculate, caudodorsal nucleus
MGBd medial geniculate body, dorsal nucleus
MGBm magnocellular division of the medial geniculate body
MGBmc caudal third of the medial geniculate body, medial division
MGBv medial geniculate body, ventral nucleus
MGBvl medial geniculate body, ventrolateral nucleus
MGm medial geniculate, medial division
MGv medial geniculate, ventral division
Ms small cell part of medial division of the medial geniculate body
MUA multiple unit activity
NBM nucleus basalis of Meynert
NLL nucleus of the lateral lemniscus
NPT posterior thalamic nucleus
OV pars ovoidea of the medial geniculate body, ventral division
P posterior auditory cortical field
paAc caudal parakoniocortex
paAlt lateral parakoniocortex
paAr rostral parakoniocortex
PAF posterior auditory cortical field
paI parainsular cortex
PHA-L phaseolus vulgaris leucoagglutinin
PL pars lateralis of the medial geniculate body
Po posterior thalamic nuclei
Pol (POl) lateral part of the posterior thalamus
PPA peripeduncular area
PPA peripeduncular area
Pro proisocortex
proA auditory prokoniocortex
PtA parietal association cortex
RE reticular nucleus of the thalamus
reIt retroinsular temporal cortex
RN reticular nucleus
SAG nucleus sagulum
SCd deep layers of the superior colliculus
sd superficial dorsal nucleus of the medial geniculate, dorsal division
SF suprasylvian fringe auditory area
SFF suprasylvian fringe sector
sg supergeniculate nucleus
SG suprageniculatum
T temporal auditory field
Te1 cortical temporal area 1
Te1v cortical temporal area 1, ventral
Te2 cortical temporal area 2
Te2c cortical temporal area 2, caudal
Te2d cortical temporal area 2, dorsal
Te3 cortical temporal area 3
Te3r cortical temporal area 3, rostral
Te3v cortical temporal area 3, ventral
TeA temporal association cortex
TGNS Theory of Group Neuronal Selection
Tpt temproparietal cortex
Ts1 temporalis superior cortex 1
Ts2 temporalis superior cortex 2
Ts3 temporalis superior cortex 3
UCS unconditioned stimulus
V (VN) ventral nucleus of the medial geniculate
VCN ventral division of the cochlear nucleus
VE ventral ectosylvian auditory cortical field
VL (vl) ventrolateral nucleus of the ventral division of the MGB
VP ventral posterior auditory cortical field
Vt transitional zone of the ventral division of the MGB
WGA-HRP wheat germ agglutinin conjugated to horseradish peroxidase
ZM marginal zone of the medial geniculate body
The terms top-down and bottom-up have been used differently in different places in the literature reviewed in this paper. Koch and Davis (1994) use the terms to describe the direction of the theoretical approach to understanding the brain. Here, top-down refers to psychological or computational approaches to brain and behavior with a minimum of concern for the underlying physical mechanisms responsible or the biological plausibility of the theory in general. The bottom-up approach requires the development of theories of brain function based on an actual and detailed biological description of the system.
The other important usage of the terms "top-down" and "bottom-up" refers to the direction of neural signaling. They are generally equivalent to the terms "efferent" and "afferent", respectively. However, an exact equivalence between these sets of terms is inaccurate. Conventional usage would lead to the association of afferent signaling with sensory information and efferent signaling with motor or endocrine control. This is consistent with the paradigm that has dominated neuroscience almost up to the present. This usage conflicts, however, with the observation of descending sensory projections from the cortex and thalamus that may equal or exceed in number the corresponding ascending projections (Winer, 1992).
There is a newly evolving paradigm of perception and cognition (Cauller, 1995; Kolb and Whishaw, 1996). This interactive paradigm of sensory processing adds a parallel top-down information flow to the bottom-up model, reflecting the biological reality of widespread, reciprocal connectivity in the brain (Edelman, 1978; Friedman, 1983; Felleman and van Essen, 1991; Pandya, 1995; Cauller, 1995). In this paper, the terms "top-down" and "bottom-up" will be used consistent with the development of this new paradigm (Cauller, 1995). The paradigm explicitly acknowledges the presence of reciprocal connectivity in the sensory system and higher cortical areas. It rejects a strictly serial, one-directional arrangement in the cortex and consequently must deal with ambiguity as to what is actually top-down and what is bottom-up because the neural activity that ascends from peripheral sensory structures always is balanced from the top-down and is therefore never purely bottom-up.
1.4 Interactive Paradigm of Sensory Processing
While at the business of clarifying terms, it is also noteworthy that the term interactive, as used in the interactive paradigm of sensory processing, has a double entendre. Interactivity can be seen in relation to the organism and its environment, as the organism engages its sensory system in actively probing its environment using the interactivity of its sensory and motor systems. In the second sense of the term, interactive refers to the interactivity between bottom-up and top-down neural signaling. It is this second meaning that will be primarily of interest in this review of the reciprocity of neural connectivity in the auditory forebrain.
Recent advances in neuroscience research have led to the modification of the traditional, hierarchical, "bottom-up" description of sensory processing (Cauller, 1995). While bottom-up information processing has been able to explain some aspects of perception, such as feature detection, its general reliance on experimental designs that utilize preparations that are anesthetized or restrained and presented artificial stimuli under unnatural behavioral conditions has led to a description of sensory perception as a passive event. Perception under natural conditions is not a passive process. Organisms actively probe their environments. This is clearly seen in the case of saccadic eye movements to scan the visual environment and the active probing of the hands and fingers in acquiring somatosensory information (Churchland et al., 1994; Cauller, 1995). Further reflection reveals that taste, smell and hearing are also active processes. Active hearing is commonly referred to as listening. It is a conscious activity that allows the listener to selectively focus on a given part of the auditory environment. The fact that listening is a behavior that is under conscious control in humans, implies that it is a neocortical process because neocortical lesions have the most direct affect upon conscious function (Luria, 1973). The interactive paradigm of sensory processing provides a framework for the study of the neural process of active listening and selective auditory discrimination.
An important problem unresolved by bottom-up processing involves integrating multiple parallel streams of sensory inputs into a unified conscious experience. A strictly bottom-up approach has proven inadequate to account for the perceptual integration of sensory information based on a neural representation of objects without resorting to the use of a homunculus or grandmother cell (Barlow, 1972). This problem, termed the binding problem, is provided a possible solution in the way that cortical areas are interconnected by top-down projections converging in the superficial layer of sensory cortex. The resolution of the binding problem may be fundamental in developing an explanation of the neural basis of consciousness (Crick and Koch, 1990).
This leads to another very important issue addressed by the interactive paradigm of sensory processing; the signaling of conscious sensation. Studies performed in the somatosensory system of monkeys (Kulics et al., 1977; Kulics, 1982) demonstrate that the N1 component of the somatosensory evoked potential in primary somatosensory cortex corresponds to the conscious discrimination of touch sensation (Cauller and Kulics, 1991). The N1 component is putatively believed to be generated by top-down projections from higher order somatosensory areas. These results provide evidence of the importance of top-down projections terminating on cortical layer I in the discrimination of conscious sensation.
The question of the mechanisms by which perceptual binding and conscious sensory discrimination operate lead to specific research questions. To date all work on these questions has been performed on the somatosensory and visual modalities. The study of how these questions relate to the auditory system would make a significant contribution towards the goal of relating these ideas to sensation in general and finding commonalties in the processing of information between the various sensory modalities. Additionally, hearing is a particularly interesting modality in humans because of its importance in our acquisition and expression of language.
The Theory of Neuronal Group Selection (TNGS) by Edelman (1987, 1989) was developed to provide an alternative to the functionalist (or information processing) approach to neuroscience. It seeks to provide an explanation for the processes of memory, generalization, and perceptual categorization in terms of neural activity. This explanation is based on the mechanisms of experiential selection, developmental selection, and reentrant mapping. The third of these, reentrant mapping, is of central interest to this literature review.
The parallel and recursive flow of information between areas of the brain forms a dynamic process of reentry (Edelman, 1989). The basic idea behind cortical reentry is that there exist multiple cortical maps in the brain, each creating a representation of some features of the stimuli in the world that they map. At least one of these maps must be topographically arranged to correspond to the sensory receptor sheets at the periphery. Other maps may be connected by reentrant projections to the topographically organized map, allowing the integration of processing of different aspects of perception and the correlation of various features of stimuli that derive from common locations on a receptor sheet.
I must point out that while this description fits well with the visual and somatosensory systems, it does not account for the perception of auditory spatial location. According to Middlebrooks (1997), there is no topographical mapping of auditory space in the auditory cortex. Auditory space must be invented by the central nervous system because the peripheral receptor sheet is not arranged to map sound as a spatial dimension. Nevertheless, while not as straight forward as vision and touch, reentry may be able to provide an explanation as to how the auditory system creates a sense of auditory space. Aspects of auditory perception such as this provide interesting challenges to the field of neuroscience.
If reentry is the process of temporally ongoing and parallel neural firing between ordered, anatomically connected, cortical maps, then we may consider the possible temporal characteristics and types of anatomical connections. Temporally, reentrant connections between any two areas may be asynchronous, synchronous and phasic, asynchronous and phasic or synchronous and cyclic. Each pair of connected areas in a reentrant network may possess any one of these temporal sets of properties. Similarly, the anatomical connections between two areas may exhibit different combinations of features. Spatially, fiber patterns may be registered, convergent or divergent. Terminal distributions may be layer registered or not, and arborized or non-arborized. As illustrated in figure 1, these features may occur in different combinations. This variety of anatomical possibilities multiplied by the possible temporal characteristics results in a large number of theoretically possible arrangements of reentrant signals. Edelman claims, without providing specific examples, that expressions of each of these combinations can found in rich nervous systems.
A list of possible functions of reentry is given in table 1. Of these, the issues of association, correlation, conflict resolution, cross-modal construction and recursive synthesis seem particularly relevant to the problems of sensory binding, conscious discrimination of sensation and the larger question of the neurobiological basis of consciousness as defined by Uttal (1978), Edelman (1978, 1987, 1989), and Crick and Koch (1990). The interaction between reciprocally or triangularly connected areas allows for integration through the correlation of activity leading to the production of globally coherent patterns. These patterns form the putative basis for the binding together of sensory information to create the experience of a unified percept.
Table 1. Some functions of reentry.
(Reproduced from Edelman, 1989.)
Reentry as discussed by Edelman (1978, 1987, 1989) and the reciprocal connectivity provided by top-down cortical projections (Cauller, 1995, Clancy, 1996) which form the anatomical basis for interactive paradigm of sensory processing are very similar if not identical. The greatest difference between the Edelman’s reentry and the Cauller’s interactive paradigm of sensory processing is in the approach. Edelman sets out to develop a theoretical explanation of brain function that is biologically plausible, while the interactive paradigm of sensory processing is an empirically based set of ideas that is closely tied to carefully documented anatomical features and experimental electrophysiological evidence. Thus distinguished, the TNGS provides a rich set of ideas that can be explored and tested, and can potentially contribute to the current shift in the neuroscience paradigm of sensory processing.
Figure 1. Some anatomical patterns showing reentrant connectivity.
(Taken from Edelman, 1989)
1.6 Reciprocal Connectivity and Hierarchical Processing
The bottom-up, serial model of sensory processing is not consistent with recent advances in knowledge of the anatomy and physiology of the brain. This model, originating with work of Hubel and Wiesel (1962,1965) provides a description of sensory processing where signals are fed-forward through a series of hierarchical levels with increasingly complex informational properties. This model has been very useful in explaining the increased complexity of cell properties along sensory pathways from the periphery to the cortex. However, it fails to take into account the reality of extensive top-down sensory projections, reentrant connectivity, and existence of parallel and interconnected sensory pathways. A review of connectivity patterns among 32 areas of visual cortex revealed a high degree of connectivity, with each area linked to other areas by an average of 19 pathways (Fellman and Van Essen, 1991). The overall connectivity level between the areas is estimated to exceed 40% of the maximum possible connections. This level of connectivity argues strongly for the idea of highly distributed cortical processing.
The hierarchical aspect of the Hubel and Wiesel model continues to be of great value, requiring only the replacement of the idea of a serial, feed-forward flow with that of reentrant connectivity and distributed processing. The study of the hierarchical organization of cortical areas and their laminar patterns of reciprocal connectivity are closely linked by the ability to define hierarchical organization based on these patterns. Evidence of this has been provided in the auditory system by (Galaburda and Pandya, 1983), the somatosensory system (Friedman, 1983) and visual system (Wong-Riley, 1978; Rockland and Pandya, 1979). Fellman and Van Essen (1991) performed an analysis of the known connectional properties of the visual and somatosensory/motor systems in the macaque monkey to test the validity of a scheme of hierarchical organization based on anatomical criteria. Explicit in the scheme is the segregation of information flow in parallel streams with communication between the streams at specific levels in the processing hierarchy. The assignment of hierarchical relationships between cortical areas is based on the patterns of connectivity as determined by the laminar location(s) of the cell bodies of origin of their connecting pathways and the layer(s) in the receiving cortical area to which they project. Forward (bottom-up) pathways are characterized by axonal termination in the middle layers, especially layer IV. The terminations of descending (top-down) pathways generally avoid layer IV and exhibit either a bilaminar pattern with endings in both deep and superficial layers, or focused mainly on just the superficial layers. The cell bodies of origin for either top-down or bottom-up projections may form a bilaminar distribution involving deep and superficial layers. Alternatively, and more characteristically, the bottom-up pathway cell bodies are found in the supragranular layers while top-down pathways originate in the infragranular layers. In addition to ascending and descending pathways, the scheme identifies a lateral projection pattern that originates in a deep/superficial bilaminar pattern and terminates in columns that include all the cortical layers. These laminar patterns are illustrated in figure 2.
While top-down connectivity is clearly a significant anatomical feature of sensory systems, the functional implications are not really known. Various suggestions for the functional role of reentrant and top-down connectivity will be presented throughout this review.
Figure 2. Laminar patterns of cortical connectivity used for
making hierarchical assignments. Three characteristic patterns of termination
are indicated in the central column. These include preferential termination
in layer 4. There are also 3 characteristic patterns for the cells of origin
of different pathways. Bilaminar (B) patterns, shown on the right, include
approximately equal numbers of cells from superficial and deep layers (no
more than a 70%-30% split) and are found to occur with all three types
of termination pattern. Unilaminar patterns include predominately superficial-layer
inputs (S pattern) which correlate with F-type terminations, and predominately
infragranular layer inputs (I pattern), which correlate with M-terminations.
Within this general framework, a number of variations on a theme can be
encountered. Some pathways terminate primarily in superficial, but we group
them with the M pattern because they avoid layer 4. Other pathways are
quasi columnar, but do not include all layers; we classify them as a C
pattern if the labeling in layer 4 is neither heavier nor sparser than
in adjoining layers.
(Figure and caption taken from Felleman and Van Essen, 1991)
2. The Auditory Forebrain: Anatomy and Function
A general principle of science found implicitly manifest throughout the literature is the principle of providing verification of knowledge by exploring the same question with various methodologies and techniques and at different levels of analysis. In the case of neuroscience, this can be seen in the complementary and interdependent relationship between anatomical and physiological studies (e.g. Rose and Woosley,1949). In the specific case of mapping out the anatomy of auditory forebrain structures, divisions of the thalamus or cortex can be variously defined by: 1) types of cells and their dendritic morphology and the arrangement of cells including their laminar organization; 2) patterns of neural connectivity with other areas; 3) neurochemistry including histochemical, immunohistochemical, and autoradiographic techniques; 4) electrophysiological properties including tonotopic maps and temporal response properties. Each of these various techniques provides an alternative view of the boundaries of thalamic nuclei or cortical fields. To the extent that they provide coincident boundaries they reinforce the schema of anatomical differentiation between these areas, as well as providing clues to the their functional significance. To the extent that they differ, they provide challenges for current and future research to resolve.
The increased precision in the anatomical description of the brain over
the course of the present century reflects improvements in available technology.
Earlier studies depending on Golgi, Nissl and myelin staining, and studies
utilizing retrograde degeneration, have been extended by the use of electronmicroscopy
and by uptake and transport studies using WGA-HRP, antisera, and florescent
dyes.
2.2 The Medial Geniculate Body
The medial geniculate body (MGB) is a set of diencephalic nuclei of the ventral thalamus which serve as relay centers for all auditory inputs to the cortex. This status as an obligatory synaptic connection is true for all mammals as well as most other vertebrates (see Winer, 1985a, 1992; Imig and Morel, 1988 for reviews). Clerici and Coleman (1990) make a point of characterizing the MGB as being more than a simple relay station for afferent information, as is exhibited by the differences in architecture, physiology and connections of the complex of nuclei that comprise it. Along the same lines, Winer (1992) points out that the differences in tuning curve properties and patterns of brain stem inputs among MG subdivisions reveal that the different MG divisions are not all equal in the degree to which they are auditory in nature. While the MGv has a cochleotopic mapping and many of the features of a synaptic relay, the MGm and MGd receive many more afferent inputs than the MGv. These nuclei receive a convergence of of inputs, both auditory and non-auditory. This convergence forms a representation with an unknown internal order (Winer, 1992).
In general, a review of the literature reveals that while many factual details of anatomy and physiology are known about the MGB, auditory neuroscience is still far short of providing an explanation of its role in audition, especially as it relates to behavior.
2.2.1 Organization of Medial Geniculate
The anatomy of the MG has been studied in greatest detail in the cat (Rose and Woosley, 1949; Morest, 1964, 1965; Winer et al., 1977; Andersen et al., 1980; Imig and Morel, 1983, 1988; Winer and Morest, 1983, Winer,1984a, b, c, d, 1985, 1992). Studies of other species include the guinea pig (Redies et al., 1989), humans (Winer, 1984), monkeys (Burton and Jones, 1976; Yeterian and Pandya, 1989, Hashikawa et al., 1995), the mustached bat (Winer and Wenstrup, 1994), the opossum (Kudo et al., 1986; Winer et al.,1988), and the rat (Winer and Larue, 1987; Scheel, 1988; Clerici and Coleman, 1990).
This review will focus on studies performed on the cat, because it is by far the most thoroughly studied species, and on the rat, because it is of the greatest interest to myself as a result of its utility and availability for study. Care must be taken in comparing different species not to overextend similarities in drawing parallels between them. Similar structures may be homologous, deriving from common origins but possessing different functions, or they may be analogous and serve the same function but differ in their embryological origins. The following description pertains to the cat except where otherwise explicitly noted.
In general, the ascending auditory pathways can be divided into a primary, lemniscal pathway retaining the tonotopic organization of the cochlea and a parallel non-lemniscal or lemniscal adjunct pathway that lacks the more tonotopic precise organization of the primary pathway (Andersen et al., 1980; see for reviews, Weinberger and Diamond, 1988; Winer, 1992). The lemniscal pathway extends from the cochlea through the ventral cochlear nucleus, the lateral lemniscus, and the central nucleus of the inferior colliculus (ICC; Beyerl, 1978) to the ventral division of the MGB (see figure 3). The MGv is tonotopically organized, receiving the majority of its afferent input from the ICC (Rouiller and Ribaupierre, 1985). In addition to the MGB, the lateral division of the posterior thalamic group is discussed as part of the auditory thalamus by some authors (Winer et al., 1977; Andersen et al, 1980; Imig and Morel, 1983, 1985, 1988) and assigned to the dorsal division of the MGB by Winer (1992).
One of the earliest modern anatomists, Ramon y Cajal (1911) proposed a tripartite scheme for the division of the MG. Later studies (Rioch, 1929; Rose and Woosley, 1949) divided the MG into two divisions, the pars principalis and the pars magnocellularis. Rose and Woosley commented that the cellular structure of the principle division does not differ significantly across that division but that cells were somewhat larger and less densely packed in its dorsal extent.
More recently, a widely accepted cytoarchitectonic scheme devised by Morest (1964) divides the MGB into three main divisions, the ventral (MGv), dorsal (MGd) and medial (MGm). Whereas Rose and Woosley saw little cellular variation across the pars principalis, Morest was able to make distinctions based largely on the differences in morphology of dendrites and of afferent plexuses of the principle (Golgi type I) neurons. The ventral and dorsal divisions can in turn be further refined into constituent subdivisions. The MGv makes up half of the total MGB and can be divided into a pars ovoidea (OV) and a pars lateralis (LV) based on the orientation of the dendritic neuropil. Additionally, Morest describes a ventro-lateral nucleus, which is distinguished by its having less tufted dendrites than those in the rest of the ventral division with which it is associated. Similarly, Morest describes three specific areas that are found in the MGd, defined on the basis of morphological considerations: the superficial dorsal nucleus, the deep dorsal nucleus, and the suprageniculate nucleus. Lastly, there is the marginal zone which is designated as a strip located external to the MGB but closely related to it by virtue of its cytoarchitecture and connectivity.
As noted by Imig and Morel (1983), Morest’s schema has provided a standardized frame of reference for much of the electrophysiology and anatomical work published since. However, it has not gone without considerable revision by various authors including Imig and Morel (see below).
Details of the morphology and laminar organization of cells in the MGv and MGd are beyond the scope of this paper (Winer, 1985a; for review see Winer, 1992). The third major division of the MGB, the medial, medial geniculate (MGm), will be discussed in detail in the following section (2.2.2).
As pointed out in section 2.1 above, it is important to confirm the veracity of anatomical divisions based on multiple methodological considerations. Much of the work in recent years has used tonotopic mapping, neural tracing or their combination in order to establish functional anatomical arrangements.
The areal boundaries of the respective nuclei of the MGB can be discriminated on the basis of their topographically organized best frequency mapping obtained from responses to peripheral stimulation. Anne Morel and Thomas Imig have published several papers that have variously parceled the auditory thalamus by different arrangements. Imig and Morel (1983) divided the auditory thalamus into five areas: the laminated ventral nucleus; the posterior group; the periventral nuclei; the deep dorsal nucleus; and medial division (see Fig. 3). Morel and Imig (1987) divided the auditory thalamus into seven nuclei. Three nuclei are described as tonotopic: the ventral nucleus (V); the lateral part of the posterior group (Po); and the dorsal cap nucleus (d). The other four are nontonotopic: the caudodorsal (cd); the ventrolateral (vl); and the small (Ms) and medium-large (Mg) cell regions of the medial division. Imig and Morel (1988) describe the auditory thalamus as consisting of four major divisions: the ventral nucleus (VN); the lateral part of posterior group (Po); the dorsal (d) and ventral (v) areas; and the medial division (M). The description of a medial division is a consistent feature of the various schemes, though in the 1988 version it is divided into two subregions according to predominant cell size. The periventral nucleus of the 1983 paper is replaced by areas cd and vl in 1987 and then becomes the d and v of the 1988 scheme. This sort of multiple nomenclature is found throughout the literature of neuroanatomy and creates for readers a source of difficulty and confusion in interpreting and comparing results.
The different divisions of the MGB are distinctive in their connectivity to the various cortical areas. For example, the MGv projects to the primary auditory cortical field (AI) and to the anterior auditory cortical field (AAF) and receives fibers primarily from AI and AAF but also receives light projections from the secondary auditory cortical field (AII). In contrast, the MGd projects to AAF but not to AI and does not receive input from either one, but is reciprocally connected to AII (Winer, 1985b, 1992). Thus, an examination of the neural connectivity of the MGB provides a complementary analysis of its anatomical organization to those provided by it cytology and morphology. A detailed review of neural connectivity is found in section 2.4.
Figure 3. Synthesis of the major components of the thalamocortical
auditory system, with special emphasis on the connectional relations between
the thalamus, cortex, and forebrain as well as midbrain and selected brain
stem centers.
(Figure and partial caption taken from Winer, 1992)
The MGB of the rat also has been divided into dorsal, ventral and medial
divisions along the pattern proposed by Morest for the cat (Ryugo and Killackey,
1974; Paxinos and Watson, 1982; Winer and Larue, 1987; Scheel, 1988).
Ryugo and Killackey (1974) utilized stereotaxically placed lesions in the MGB and the Fink-Heimer histological procedure to examine the differential thalamocortical projections from the MGv and MGm. The results showed that these two divisions could be differentiated by their patterns of fiber degeneration. The study is significant in that it did provide clear evidence to support the parcellation of MGv and MGm in the rat along the lines of Morest. Details on the patterns of connectivity found and discussion of results is found in section 2.4.1 below.
Patterson (1977, dissertation) conducted a comprehensive study of the organization of the rat MGB and identified medial, ventral and dorsal divisions as well as a ventro-lateral nucleus, a marginal zone and a supergeniculate nucleus. He also identified a caudal division occupying the posterior extremity of the MGB, replacing part of Morest’s MGd. The cytoarchitecture and myeloarchitecture were examined using cresyl-violet and silver stains and a study of the thalamocortical projections from the MGB was performed by means of anterograde degeneration and the retrograde transport of horseradish peroxidase (HRP). The results provide detailed support for the extension of Morest’s findings (see above) to the rat. Details pertaining to the MGm will be discussed in section 2.2.2 and findings relating to connectivity in section 2.4.1.
In her 1985 study of cortical projections to the inferior colliculus, Faye-Lund wrote that the rat MGB had yet to be the subject of a detailed cytoarchitectonic study to provide guidelines for its parcellation in the current study, apparently overlooking the unpublished work of Patterson (1977). In the next few years, several researchers (Winer and Larue, 1987; Scheel, 1988; Roger and Arnault, 1989; Clerici and Coleman, 1990) published the results of studies which contribute to a detailed description of the organization of the rat auditory thalamus. These papers are briefly reviewed below.
The paper by Winer and Larue (1987) is a extensive and significant piece of work that covers a number of research questions relating to reciprocal connectivity between the auditory thalamus and cortex in the rat. Discussion in this section will be limited to its attempt to provide a description of the organization the MGB in rat. A more detailed consideration their findings on the MGm is found in the next section (2.3.1); the remainder is discussed in the section on neural connectivity (2.4.1).
Winer and Larue divided the rat MGB into the same three divisions as did Morest: ventral, dorsal and medial. Their division was based on myeloarchitecture, cytoarchitecture and the connectivity between the thalamus and cortex. The ventral division was described as containing mainly medium diameter (12-15 m m) fusiform cells and smaller(<10m m) oval or round cells. The fusiform cells had tufts of bushy dendritic branching at the ventral and dorsal poles. The smaller cells were lacking in dendritic tufts, having an irregular, radiate pattern of shorter, smoother and thinner dendritic processes than the fusiform cells. The ventral division is characterized by a distinct pattern of fiber architecture. Fibers course parallel to the orientation of the dendritic tufts of the fusiform cells, creating an ventromedial-dorsolateral aligned plexus. A description and illustration is given of the pars ovoidea region, occupying the medial aspect of the ventral division with its concentrically arranged fibers swirling around in the face of the brachium of the inferior colliculus. The description given of the dorsal division is that of a reticulated arrangement of clusters of 5-20 cells interspersed with fibers. The somata are mostly round or oval and are generally larger than those found in the ventral division, ranging up to 20 m m; present also are a few smaller cells (<10 m m). The myeloarchitecture differs significantly from that of the ventral division. Rather than a ordered, laminar arrangement, the fiber plexus consisted of smaller-gage axons forming a dense, patternless mesh.
Scheel (1988) performed a study of the topographic organization of rat auditory thalamocortical connectivity by means of retrograde tracing. The study described the rat auditory thalamus as comprising the following areas: the posterior thalamic nucleus (NPT); and the medial geniculate body (MGB) containing the medial (MGBm), lateral (MGBv), dorsal (MGBd) divisions, the ventrolateral nucleus (MGBvl), the caudal third of the MGBm (MGBmc), and the suprageniculatum (SG). The source of this divisional arrangement was a study by Imig and Morel (1985). The results of the HRP-WGA transport lend support for a general tripartite schema in the rat, while providing a more refined analysis with the additional description of the nuclei listed above. Aspects of this study relating to cortical organization (section 2.3) and thalamocortical connectivity (section 2.4.1) are discussed below.
Roger and Arnault (1989) examined the connectivity of the primary auditory cortex in rat with emphasis on the topographical organization of thalamocortical projections. They followed this up with a study (Arnault and Roger, 1990) of the thalamocortical connectivity of the secondary auditory cortex. These studies used the nomenclature from Morest (1964) and identified five major areas of auditory thalamus based on cytoarchitectural considerations, with some additional subdivisions. They are: the ventral nucleus (MGv); the dorsal nucleus (MGd), subdivided into the caudodorsal nucleus (cd), deep dorsal (dd), dorsal (d), and superficial dorsal (sd);the medial nucleus (MGm); the supergeniculate nucleus(Sg); the lateral part of the posterior thalamus (Pol);and the peripeduncular area (PPA). These two studies are primarily concerned with the definition of cortical areas and cortical connectivity (discussed below) but they provide additional information to support the application of Morest’s classification scheme to the rat with the addition of the criteria of connectivity to that of cytoarchitecture.
The studies of the rat MGB described above, all lacked detailed definition
of nuclear boundaries and did not utilize an analysis of dendritic morphology
for verification as did Morest’s studies in the cat. Clerici and Coleman
(1990) performed a detailed study of the morphology of neurons in the rat
MG and provided verification of the extension of Morest’s organization
to the rat. They examined cytoarchitecture, myeloarchitecture, dendritic
morphology and cortical connectivity to provide precise boundaries supported
by multiple methodologies. The details of the study are beyond the scope
of this review, but the basic finding strongly supports the division of
rat auditory thalamus along the same lines as provided by Morest (1964)
for the cat. The paper provides support for a general scheme of mammalian
auditory thalamic organization and provides details of its extension in
the rat.
2.2.2 Medial Division of the Medial Geniculate
The medial division of the medial geniculate body (MGm) is of special interest in this review of the auditory forebrain due to its special characteristics and unique pattern of connectivity. An important question to be resolved by future research is the role of reentrant processing provided by the reciprocal connections between the auditory cortex and the medial division of the medial geniculate body. The following overview of papers provides a picture of the present level of understanding of the anatomy and physiology of the MGm.
As discussed above in section 2.2, Morest performed studies of the division of the MGB based on neuronal morphology (1964). Morest’s description of the medial division is relatively brief. In his discussion section, he points out some of the limitations of the study pertaining to the MGm. He suggests that there may be more neuronal populations in the MGm than were identified at the time and that further analysis would be required before statements about the function significance of the division could made.
In 1983, Winer and Morest made a detailed study of the MGm. They delineated the MGm based on the criteria of its low packing density, its reticulated appearance, the distinct cytoarchitecture and myeloarchitecture. They examined the neuronal architecture of cells in the area identified as MGm and reported that it was made up of numerous cell types varying in size and shape. There are two general types of principle neurons in the thalamus: stellate and bushy cells. The degree of expression of these characteristic shapes varies. In the MGm moderate radiate branching of stellate cells is typically expressed along with intermediate forms that fall between the stellate and bushy patterns. The authors provide descriptions of five types of cells. The classification of cell types are based on the size of soma and on the configuration of their dendrites. The following are described in detail: medium-sized stellate neuron; medium-sized tufted cell; medium-sized elongate neuron; magnocellular neuron; and the local circuit neuron. Also detailed are the variety of axonal types found in MGm. The summary finding of the study is the considerable variety of cell and axon types in the MGm. This heterogeneity is one feature that sets the MGm apart from the rest of the MGB. Another distinctive feature is the apparent lack of subdivision within the MGm. The various cell types are mixed together throughout the MGm, though regional differences exist as seen in a higher proportion of stellate cells in the caudal aspect and more bushy and magnocellular neurons located rostrally. Given the variety of neurons, the term pars magnocellularis applied to the MGm is a misnomer in that the magnocellular neurons comprise only a small minority its cells. The morphological diversity and pattern of connectivity has implications for discerning the function of the MGm. The MGm receives a range of afferent inputs other than auditory and efferent projections from non-auditory as well as all areas of auditory cortex. Its polymodal nature as a result of this diversity points to possible functions other than strictly auditory and its general morphological diversity suggests that it may have multiple functions and not act as a singular unit.
A more extensive consideration of possible function of the MGm comes from the electrophysiological studies of cortical and thalamic neural plasticity of Norman Weinberger and colleagues (Edeline and Weinberger, 1992; Gerren and Weinberger, 1983;
Ryugo and Weinberger, 1978; Weinberger, 1993,1995; Weinberger and Diamond, 1987).
Weinberger and Diamond (1987) provide a extensive review of the study of properties of physiological plasticity in the auditory forebrain. Aspects relevant to the MGm are discussed here, other aspects are reviewed elsewhere in this paper.
The MGm is the only portion of the auditory thalamus that receives a convergence of afferent input from the midbrain via both the lemniscal line and lemniscal adjunct pathways (see figure 4). This provides a mixture of both tonotopic and non-tonotopic inputs. Additionally, neurons in the MGm are responsive to somatosensory stimuli. It also differs from the rest of the medial geniculate body in its unique pattern of reciprocal connections with the auditory cortex. Unlike other areas of the MG, which project to the middle cortical layers, the MGm projects primarily to layer I of all areas of the auditory cortex. It receives its cortical inputs from fibers which originate in cortical layer V, unlike the rest of the MG, which receives inputs from cells in layer VI. These anatomical findings may relate to the electrophysiological finding that it is the only area of the auditory thalamus that exhibits the property of enduring learning-induced physiological plasticity (Ryugo and Weinberger, 1978) and are discussed below in relation to information storage and retrieval in auditory cortex (Edeline and Weinberger, 1992). The issue of whether physiological plasticity observed in the MGm is produced locally or may be derived elsewhere is addressed by the demonstration of locally produced synaptic plasticity. Long-term potentiation was produced in the MGm by high frequency stimulation of the brachium of the inferior colliculus (Gerren and Weinberger, 1983).
Figure 4. Schematic diagram of the organization of higher order
pathways in the auditory system. The lemniscal line is striped left, the
diffuse pathway is solid and the lemniscal adjunct in striped right. According
to this schema, cortical and midbrain regions may be components of more
than one pathway, e.g. Icc, AAF and AI are connected with both, the thalamic
divisions of the lemniscal line (MGv) and diffuse pathway (MGm); Icx and
AII are connected with the diffuse and lemniscal adjunct thalamic nuclei;
and PAF, VPAF and VE are terminal fields. See Abbreviation Table for nomenclature.
(Figure and caption taken from Weinberger and Diamond, 1987)
Edeline and Weinberger, (1992) found receptive field plasticity produced in the MGm by classical conditioning. The plasticity was frequency specific to the conditioned stimulus frequency. Details of the results will not be covered here, but major points of interest from the discussion will now be summarized.
If plasticity is found in both the thalamus and connected cortical areas, it is possible to assume that this property is projected to the cortex by the thalamus. However, the reverse may also be true, the property may obtain in the cortex and be projected onto the thalamus. As was pointed out by Winer (1992), there are many times the number of corticothalamic fibers as there are thalamocortical ones. The experimental evidence presented indicates that the physiological plasticity in the medial geniculate cannot fully account for the plasticity observed in the cortex.
The authors hypothesize that the auditory cortex, MGv and MGm operate as an integrated whole instead of independent, serial, feedforward pathways. Thalamic and cortical plasticity are viewed as being interactively related. The integrated cortical connectivity of the auditory forebrain gives rise to a model of information storage and representation based on associatively produced plasticity. The plasticity of the MGm may provide a mechanism for the retrieval of stored information from the auditory cortex.
The model is based on learning during classical conditioning trials. With paired conditioned (CS) and unconditioned (UCS) stimuli, the tonotopically organized MGv relays accurate frequency information (the CS) to the layer IV of the auditory cortex. The MGm receives input from both the CS and UCS and exhibits associative plasticity. The MGm does not have the narrow tonotopic tuning of the MGv. Instead, it relays a general increase in response as a function of exposure to the CS with layer I as its laminar target. Simultaneously, the MGm projects to the amygdala, which in turn projects to the nucleus basalis of Meynert (NBM), stimulating the release of acetylcholine (Ach) onto the apical dendrites of pyramidal cells in the cortex. The acetylcholine provides a postsynaptic enhancement and paired with the MGv input results in a Hebbian type learning for the CS frequency. Further discussion of the role of acetylcholine in learning-induced cortical plasticity is found in Weinberger (1993,1995) and most recently in a study by Kilgard and Merzenich (1998) that demonstrates a massive remapping of the auditory cortex in the rat as a result of stimulation of the NBM paired with the presentation of a tone stimulus.
The convergent input of the CS and UCS allows for the associative store of the acoustical features of the auditory stimulus (the CS) and the behavioral relevance of the UCS. The putative role of MGm plasticity as a retrieval mechanism is based on the previously described model of information storage combined with features specific to the MGm. The ability of the MGm to rapidly develop and retain CS-frequency specific plasticity is fundamental to its ability to retrieve information. The CS specific receptive field plasticity provides the accuracy to allow for specificity of recall. The above features form the substrate for the retrieval mechanism; the access to information stored in the cortex derives from the pattern of thalamocortical connectivity. As noted above, the MGm has a unique pattern of connectivity. The MGm projects diffusely to all of the auditory cortical areas with its inputs focused on layer I (see also Burton and Jones, 1976). The projection to all auditory cortical fields provides the means to bind the various representational aspects that unify to create a memory. The pattern of cortical cells activated by the MGm during learning form a network that is the representation of the information learned. Later exposure to the CS would reactivate the network through facilitated responses engaged by the layer I inputs and by definition result in the recall of the stored memory. The MGm need not act alone in producing the network reactivation. Even though the MGv lacks the property of retention of physiological plasticity, the concurrent input of frequency specific CS information to layer IV could act to increase the activation of the network (see figure 5). The authors suggest that it is the integration of the frequency specificity of the lemniscal path with the learned content of the non-lemniscal path that acts to store and retrieve auditory memories. The potential significance of these ideas in relation to the interactive paradigm of sensory processing will be discussed in section 3.0.
Figure 5. Diagram showing the major components of a model of
conditioning-induced receptive field plasticity in the auditory cortex
and initiation of behavior conditioned responses. The mechanisms for associative
CS-specific receptive field plasticity in the auditory cortex are based
on the convergence of three subcortical systems at the auditory cortex,
which provide detailed frequency information (‘lemniscal non-plastic’),
indicate the behavioral significance of a current acoustic stimulus (‘non-lemniscal
plastic’) and exemplify Ach neuromodulation of pyramidal cells based on
the importance of the current auditory stimulus (‘modulatory’).
(Figure and caption taken from Weinberger 1993; )
Rouiller et al. (1989) performed a study of the functional organization of the MGm based on the response properties of single cells complemented by WGA-HRP retrograde tracing. The electrophysiological results support the finding of Morel et al. (1987) that the MGm, contrary to previous belief, does have a tonotopic arrangement. The tonotopicity was more precisely defined in the anterior half of the MGm and was arranged from low to high frequency and along a latero-ventral to dorso-medial axis.
There exists a conflict in the usage of nomenclature of auditory pathways between Weinberger and Rouiller et al. What is referred to as the lemniscal adjunct pathway by Weinberger is named the diffuse pathway by Rouiller et al. Weinberger uses the term diffuse pathway to refer to the connectivity of the MGm (see figure 4). Further usage of these terms in this review will conform to that of Weinberger and Diamond (1987).
Injections of WGA-HRP into physiologically identified areas of auditory cortex showed the MGm to project to both tonotopically arranged (AI,AAF,PAF) as well as the non-tonotopic cortex (AII). The results from the tonotopically mapped areas of cortex, AI and AAF, showed labeled cells in the MGm according to the same pattern detected by single cell recordings, low to high, latero-ventrally to dorso-medially. The percentage contributions of thalamocortical fibers from the MGm to specific cortical fields were estimated as: AI,10-18%; AII, 10%; PAF, 20%; and AAF, 30%.
Along the same lines as discussed by Weinberger and Diamond (1987) the authors suggest that the MGm may act as an interface between the lemniscal and lemniscal adjunct pathways. They propose a functional role as a "comparator" of the distinct information carried by these separate pathways. No details are given as to how this might work but in addition to comparing the parallel inputs of auditory it is also pointed out that the MGm receives a variety of non-auditory inputs including somatosensory and vestibular.
The study is noteworthy as one of the few works to focus on the organization of the MGm, providing verification of the tonotopic organization of the MGm and additional details of its arrangement.
Patterson (1977) provides a brief description of the histological characteristics of the medial division in the rat. The MGm is identified using both Nissl and myelin stains. The large cells in the MGm are similar in size to the large cells in the MGd. However, with Nissl stain, the large cells in the MGm stain more darkly than any other cells in the MGB. Smaller cells are interspersed with these large cells with the ratio of large to small cells decreasing in the antero-dorsal aspect of the MGm. The fibers of the BIC cross through the MGm and provide for intense staining with myelin stain, distinguishing the MGm from the ventral division. The MGm is positioned medial to the ventral division. It extends posteriorly from the BIC to 0.75mm anterior to the brachium of the superior colliculus. The utility of the study’s atlas is limited by the poor quality of the figures and the failure to label sections with distances to a standard reference such as bregma. In spite of the limited description of the rat MGm and the shortcomings of its figures, Patterson’s work is significant as it represents the earliest attempt found in the literature to parcel the rat MGB on the basis of cytoarchitecture, myeloarchitecture and thalamocortical connectivity.
Clerici and Coleman (1990) performed a detailed study of the cytoarchitecture, myeloarchitecture and connectivity of the MGB of the rat (see section 2.2.1). In addition to confirming the validity of Morest’s schema to the rat, the study is valuable for providing a description of the internal organization of the nuclear areas. Their description of the rat MGm is clearer and more detailed than that given by Patterson, and is summarized below.
The rat MGm shares many of the characteristics found in the cat including
a reticulated appearance derived from fibers crossing from the brachium
of the inferior colliculus, and heterogeneity in the types, sizes and packing
densities of neurons found there. Unlike the cat, the magnocellular neurons
are found predominantly in the caudal rather than the rostral aspect of
the division. The lateral nuclear border of the MGm with the MGv is easily
identified by the cytoarchitectural transition expressed there. The medial
and rostral borders are made difficult to delineate due to the large number
of fibers that cross throughout these areas. The MGm has the widest variety
of cell sizes and the least uniform packing density of the MGB. Found mostly
in the caudal extent, the magnocellular neurons mostly have oval, elongated
or triangular somas which stain dark, but they may also be round and pale.
Cell size and packing density decreases towards the rostral aspect until
the cell poor, fiberous rostral extreme is reached short of the rostral
pole of the geniculate. Cells are reported to be aligned along a rostromedial
orientation; however, no pattern was found to allow for the delineation
of intrinsic subdivisions within the MGm.
2.3 The Organization of Auditory Cortex
The development of the literature describing the organization of auditory cortex by its parcellation into cortical fields closely parallels that of the auditory thalamus. The pattern of interconnectivity between the thalamus and cortex provides anatomical evidence for establishing divisions in both the thalamus and cortex to complement that provided by cytoarchitectural and other studies. Also, as emphasized by Jones and Burton, (1976) corticothalamic connections provide a guide to cortical parcellation where the is a lack of clear demarcation based on morphological transition.
The organization and connectivity of auditory cortex has been extensively studied in the cat and monkey, with much less work having been done in the rat. Brugge and Reale (1984) provide an excellent general review of auditory cortex research. A second good general reference is a small book by Aitkin (1990) entitled The Auditory Cortex.
Next, a series of notable papers that outline the course of scientific inquiry into the organization of the auditory cortex are briefly described. The emphasis in the following section will remain on the cat and rat; however, a review article of the work of Pandya (1995) based on humans and monkeys will be discussed due to its unique relevance to anatomical aspects of the interactive paradigm of sensory processing.
The work of Rose (1949) provides a good foundation for the study of cortical organization. He begins with an extended discussion of what constitutes a cortical field and explores considerations in determining architectonic borders. The study provides a description of the division of cat auditory cortex based on its cytoarchitectonic structure. A central, primary auditory area is surrounded by a peripheral belt divisible into three subdivisions. These divisions are the suprasylvian fringe sector, the secondary auditory area, and the posterior ectosylvian area. The primary auditory area is identifiable on the basis of the dense packing of small cells with little variation in size and a blurring of the lamination of cortical layers II-IV. Fairly detailed descriptions of the cytoarchitecture of the primary and belt areas are given with emphasis on the laminar differences between areas. The belt areas are described as transition zones from the central field and surrounding cortex. Rose points out that while differences in cytoarchitectonics are the basis for division of the various fields, there is also variation within the individual fields. This ties in with his introductory discussion and makes clear the somewhat arbitrary nature of parcellation based solely on cytoarchitectonic distinctions.
In a companion paper, Rose and Woosley (1949) examined thalamocortical connections to auditory cortex in relation to the cytoarchitectonic map discussed above and also to electrophysiological data from an earlier study (Woosley and Walzl, 1942). The results from these various approaches converged to support each other in providing a division of cat auditory cortex. The study possessed the limitation of utilizing ablation of the cortex and examination of retrograde degeneration in the MGB to demonstrate cortical connectivity. Details of thalamocortical connectivity will be discussed later in section 2.4.1.
Further studies led Woosley (1960) to publish a map of cat auditory cortex based on retrograde degeneration, cytoarchitecture and electrophysiological properties. His division included four tonotopic areas: the primary (AI), secondary (AII), posterior ectosylvian (EP), suprasylvian fringe sector (SSF) and into belt areas: temporal (T), insular field (I) and association cortex. In later studies, part of Woosley’s SSF was included in the anterior auditory area defined by Knight (1977) and Reale and Imig (1980).
Winer et al. (1977) conducted a study intended to verify the generally accepted division of cat auditory cortex along the lines described by Woosley reported above and to advance the understanding of the anatomical relationship between the thalamus and cortex. They emphasize the importance of using connectional studies to complement the results obtained through cytoarchitectural and electrophysiological studies. In an effort to improve on earlier results that utilized neuronal degeneration to demonstrate cortical division based on thalamocortical connectivity, they used injections of HRP to expose patterns of connectivity between the cortical and thalamic divisions. In brief, their results indicate that each cytoarchitectural division of the cortex has a unique pattern of thalamic connections. This finding supports the idea that cortical subdivisions also most likely possess distinct functions. Further details of thalamocortical connectivity are discussed below in section 2.4.1.
Reale and Imig (1980) performed microelectrode mapping of the cat auditory cortex to characterize the spatial distribution of best frequency responses. They mapped the cortex into four tonotopically organized fields: primary auditory (AI), anterior auditory (A), posterior auditory (P), and ventroposterior (VP). The posterior and ventral posterior areas are mostly buried in the posterior ectosylvian sulcus; these areas were not mapped by Woosley, and partially correspond with his posterior ectosylvian field. Each of the areas defined by Reale and Imig was said to contain a complete and orderly representation of the range of tones audible to the cat. Best frequencies were found to be arranged along a gradient from low to high in A and P, and high to low in AI and VP, along a rostral caudal axis. Isofrequency contours, lines of similar best frequencies, were found to be aligned orthogonal to frequency gradients. In addition to the core tonotopic areas, belt areas were divided into a temporal area (T), secondary auditory area (AII), ventral area (V), and a dorsoposterior area (DP). These belt areas were not mapped out in any detail. This study is noteworthy for extending earlier work in mapping to include the posterior ectosylvian sulcus, and for the detailed spatial resolution of the mapping which revealed the pattern of frequency gradients and isofrequency contours. The study also provided the foundation for a companion study (Imig and Reale, 1980) which matches the tonotopic organization to the pattern of cortico-cortical connectivity of the frequency map defined areas.
Attention will now be given to review the organization of rat auditory cortex. The auditory cortex of the rat has been parceled into primary auditory cortex (Te1) and secondary cortices (Te2 and Te3). This division and nomenclature derives from Zilles (1985) and Zilles and Wree (1985) and is based on quantitative cytoarchitectonic studies. Te1 is described as having laminar features typical of sensory cortex, with a distinct granular layer IV, and emphasis on the development of layer V and a thickening of layer I. A more detailed version is described by Romanski and LeDoux (1993b) where a ventral subfield of Te1 is partitioned as Te1v. (see figure 14) Also, based on Zilles et al. (1990), Te2 is divided into caudal (Te2c) and dorsal (Te2d) fields and Te3 into rostral (Te3r) and ventral (Te3v) fields. Primary auditory cortex can also be distinguished from the secondary cortices by its increased overall thickness and its heavy pattern of myelination (Winer, 1982). Zilles and Wree (1985) relate their map to the earlier work of Krieg (1946a, b) and of Patterson (1977). Krieg delineated a primary auditory area, designated area 41 and two secondary areas numbered 20 and 36. These areas only roughly map onto the organization of Zilles and Wree (1985; see figure 12b). Patterson’s (1977) description of a core area corresponds well with Te1 and his belt area supports the delineation of Te2 and Te3. Most recently, Paxinos and Watson (1998) have published an updated stereotaxic atlas that departs from their earlier work. In the new schema, illustrated in figure 6, auditory cortex is divided into a primary auditory cortical field (Au1) and into secondary auditory cortex, with a dorsal division (AuD) and a ventral division (AuV). Dorsal to AuD is a parietal association area (PtA) and ventral to AuV is the temporal association area (TeA).
Figure 6. Schematic diagram of coronal section of rat brain at
the approximate rostral to caudal center of the primary auditory cortex
(Au1). The dorsal (AuD) and ventral (AuV) secondary cortical fields are
both also present at this level.
(Taken from Paxinos and Watson, 1998, CD-ROM version)
While the focus of this review has been on the auditory forebrain structure and function of the cat and rat, much work has been done in the mapping of auditory cortex in monkeys. Pandya (1995) provides a recent review of the anatomy of the primate auditory cortex with an emphasis on the accumulated results of his own research. Aspects of the parcellation of cortex into subdivisions will be briefly recounted here.
In the monkey, the auditory cortices are found in the superior temporal region with the primary auditory cortex (AI) located on the supratemporal plane and surrounded by association areas. The auditory areas can be divided into three parallel architectonic lines: the root line, core line and belt line (see figure 7a). Each of these lines is characterized by progressive laminar differentiation from the proisocortex of the temporal pole up to the parietal cortex(see figure 7b). Within each line are defined subregions based on this progressive architectonic differentiation. The root line, possessing limbic features is comprised of proisocortex (Pro), parainsular cortex (paI), auditory prokoniocortex (proA) and retroinsular temporal cortex (reIt). The belt line consists of four adjacent divisions: superior temporal cortex areas 1 and 3 (Ts1, Ts3) followed by lateral parakoniocortex (paAlt) and then temproparietal cortex (Tpt) (see figure 7c). The belt line is marked by increasing emphasis on the development of layers III and IV along its rostrocaudal axis of its ordered areas. The core line, interposed between the root and belt lines, is characterized by the increasing degree of layer IV development from area 2 of the superior temporal cortex (Ts2) through rostral parakoniocortex (paAr) to its maximum expression in koniocortex (KA) and then to its termination in caudal parakoniocortex (paAc). The superior temporal plane can be alternatively be described as be divided into four rostralcaudal stages each with a core area adjoined by a root area medially and a belt area laterally (see figure 7d). The reciprocal connectional specifications of these areas and discussion of their possible functional significance will follow in the next sections.
Figure 7. (A) Diagram of the lateral surface of the cerebral
hemisphere showing three architectonic trends in the superior temporal
region. Note the root trend in the circular sulcus, the core trend in the
supratemporal plane, and the belt trend in the superior temporal gyrus.
(B) Distribution of the architectonic trends shown in (A). (C,D) Four architectonic
stages within the superior temporal region, and the intrinsic connections
between (C) and within (D) these stages (shown by arrows).
(Figure and caption taken from Pandya and Yeterian, 1984)
Tracing neural connectivity is an important technique for defining cortical areas. It has come to be seen as a more sensitive method for partitioning cortical areas than parcellation according to apparent cytoarchitectonic differences (Felleman and Van Essen, 1991). Tracing cortical connectivity provides not only a means to define anatomical areas, but also provides additional information as to what functional significance the various divisions might have. As emphasized by Rouiller et al. (1989), Felleman and Van Essen, (1991); Edeline and Weinberger (1992) and Weinberger (1993) the processing of auditory information is probably not a serial, hierarchical process along orderly, segregated pathways, but rather an integrated, reentrant process involving connectional loops with some sort of integration of informational streams occurring at both the thalamic and cortical levels. Given that physiology follows from anatomy, the study of reciprocal cortical and thalamocortical connections is central to developing an understanding of the neural basis of auditory processing and related behavior.
Earlier researchers who used tract tracing to study the anatomy of the cortex were limited to creating localized lesions in the brain and looking for degeneration in other areas. (see review above on Rose and Woosley, 1949 and Ryugo and Killackey 1974). This method, while providing some results, is fairly crude and particularly limited in its ability to show retrograde degeneration in cells which have diffuse projections not limited to the area lesioned. Neural tracing, based on fast axonal transport allows for the visualization of neurons through the use of various compounds which are taken up and transported. Earlier techniques in neural tracing made use of radioactively labeled amino acids and sugars for anterograde transport and horseradish peroxidase for retrograde transport (Kandel, Schwartz and Jessell, 1991). Horseradish peroxidase can also be used for anterograde transport and is 20-30 times more potent when conjugated with wheat germ agglutinin (WGA). The resulting HRP-WGA has uptake and transport properties in the anterograde, retrograde, collateral and transganglionic directions (Mesulam, 1982).
Immunocytochemical techniques have also been developed. For example, the use of phaseolus vulgaris leucoagglutinin (PHA-L) as an anterograde tracer provides a very powerful method when very small injections are required (see review above of Romanski and LeDoux, 1993b).
The recent development of a variety of anterograde and retrograde florescent dyes has further expanded the horizon of possibilities in tracing neuronal connectivity. The study of reentrant connections is particularly suited to the use of multiple, distinctive dyes to demonstrate reciprocity. Several dyes may be used in a single animal to allow examination of collateral branching revealed by double labeling of cells.
Fluorescent dyes may also be custom tailored for specific uses and combined with other techniques. Hill and Oliver (1993) describe a method combining the use rhodamine labeled latex spheres and biotinylated Lucifer Yellow to provide permanent visualization of cells in the inferior colliclus of the rat that were labeled by retrograde transport from MGB. A later study (Oliver et al., 1994), used the technique to trace the synaptic organization of the auditory circuit from the cochlear nucleus to the inferior colliculus and then to the MGB.
Clancy (1996) found the retrograde tracer diamidino yellow to be more
reliable than HRP when used in surface applications to label fibers projecting
to layer I. She reported that HRP often failed to label entire sets of
neurons with known connectional properties. A similar problem was reported
by Pontes et al. (1975) when using [3H]leucine to trace
corticothalamic projections. These cases point out the importance of maintaining
vigilance to the possibility that a given technique may differentially
label various classes of neurons or only partially label certain cells.
Consistent (but incomplete) results could easily be taken at face value.
The repetition of a tracing study with several compounds to compare results
is advisable to avoid this type of error.
2.4.1 Thalamocortical and Corticothalamic Connections
There has been a significant amount of study of the fiber projections between the auditory cortex and the auditory divisions of the thalamus. However, the vast majority of these studies have focused on the pattern of projections in only one direction, from thalamus to the cortex. Many fewer papers have discussed the connections from the cortex to the thalamus. A Medline search using the keyword auditory cortex combined using the AND function with the keyword thalamocortical yielded 77 hits, whereas its combination with corticothalamic yielded only 14 hits. Only a very few authors have explicitly acknowledged the potential importance of reciprocity in the connectivity of the auditory thalamus with auditory cortex (Andersen et al., 1980; Winer and Larue, 1987; Rouiller and Welker, 1991; Bajo et al., 1995; reviews: Winer, 1992; Pandya, 1995). However, as emphasized by Winer (1992) there are greater numbers of fibers projecting from the auditory cortex to the medial geniculate body than from the MGB to the auditory cortex. A detailed understanding of the auditory system must give an accounting of this connectivity. The actual functions provided by top-down projections remain uncertain (Andersen et al., 1980; Aitkin,1990; Winer, 1992). In addition to perceptual binding, suggested roles of corticothalamic inputs include: influencing levels of attention; providing selective signal amplification/suppression; gain control; behavioral readiness; and temporal delay (Winer 1992). Discussion of reentrant information processing and behavior is found in section 3 below. Attention will now be given to the review of specific papers that report on auditory thalamocortical, corticothalamic and reciprocal connectivity. Studies conducted using the cat will be covered first, followed by those in the rat.
As discussed above, Rose and Woosley (1949) examined thalamocortical connections to auditory cortex in relation to the cytoarchitectonic map of Rose (1949) and to the electrophysiological data from (Woosley and Walzl, 1942). The removal of the auditory cortex in one hemisphere resulted in the degeneration of the principle division of the MGB. They explained this finding in terms of the first (primary) auditory area receiving "essential projections" from the principle division. Essential projections are defined as fiber terminations that if destroyed result in the degeneration of the cell bodies of origin. Focal lesions in AI revealed the principle division to project upon AI in an orderly fashion along a corresponding dorsoanterior to ventroposterior axis. The magnocellular division was spared degeneration both following ablation of the entire auditory cortex and from focal lesions in AI and the belt cortices and was therefore reported not to send essential projections to any part of the auditory cortex. Localized lesions placed in auditory belt areas did not produce retrograde degeneration in the MGB. The authors emphasize that the lack of essential projections does not equate with the lack of projections to an area. Neurons with a diffuse pattern of termination in a particular area or that send collaterals to several areas would be spared degeneration by virtue of being sustained by their remaining terminals.
The results were seen as valuable in confirming the results of Rose’s (1949) cytoarchitectonic study as well as the electrophysiological data from Woosley and Walzl (1942). The concordance from the interweaving of multiple methodologies is a strong point in the study. Specifically, the combination of functional and anatomical data allowed the firm delineation of the primary auditory area. However, due to the limitations of fiber degeneration, only negative findings were able to reported concerning the connectivity of the magnocellular division of the MGB and the belt cortices.
At the time of the study reported in Raczkowski et al. (1975), the projection of the MGv to AI had been well established along with the diffuse pattern of projection from MGm to a wide area of auditory cortex. However, the specific connectivity of the MGB to auditory cortical areas other than AI was still uncertain, in part due to the limitations of neuronal degeneration as a method of investigation. Raczkowski et al. (1975) conducted a retrograde labeling study using HRP in the cat. Restricted injections of HRP were made into auditory cortex in an effort to establish the thalamic locations of cell bodies projecting to specific auditory fields. The MGB and auditory cortex were divided in general agreement with the schemes of Morest (1964) and of Rose and Woosley (1949) respectively. They presented several cases, representing injections into AI, AII, Ins, Ep and the temporal auditory area. The injection into AI demonstrated a topographic relationship between the organization of AI and MGv with a rostral to caudal axis in AI mapped to a medial to lateral axis in MGv. The focus of labeled cells in AI formed a dense and concentrated band. Injections into AI, AII, Ins, Ep resulted in labeling in the dorsal and magnocellular regions of the MGB and the posterior thalamic group. The pattern of labeling from AI, AII, Ins, Ep was described as widely scattered and lacking the density of the AI projection to MGv. The only distinctive pattern found was that from the temporal area, it produced a concentration of labeling in the extreme caudal extent of the MGB. Larger injections, spreading over several cortical areas were reported to not increase the extent of labeling in the MG but did increase the density of cells labeled. The authors conclude that there are two patterns of connectivity between the MGB and the divisions of auditory cortex: a precise topographical projection pattern involving only MGv and AI; and a pattern of overlapping connectivity between the other cortical areas examined and the MGB, (excluding the MGv) where each thalamic area projects to several cortical areas and each cortical are receives inputs from several thalamic nuclei. A later, complete report by the same authors (Winer et al, 1977) provides a more refined description of this connectivity. The study is significant in representing an early attempt to clarify the organization of thalamocortical connectivity in the auditory system using a neural tracing technique based on fast axonal transport.
Winer et al. (1977) adds some refinement to the results described above from Raczkowski et al. (1975). As stated above, each division of the MGB (except MGv) projects to two or more auditory cortical areas. The dorsal division projects heavily to Ep with lighter labeling in AI, AII and the temporal area. The magnocellular division projects to all areas of the auditory cortex. As previously reported, the caudal pole of the dorsal division sends fibers primarily to the temporal auditory cortex, but with some projections to Ep and AII also. The medial part of the posterior thalamic nucleus projects mainly to the insular cortex and also to the posterior ectosylvian gyrus. These results supported the idea that every cytoarchitectonic subdivision of the cortex has a unique set of connections with the thalamus. This observation leads to the question of what is the functional significance of these individual cortical areas and their separate pathways. The authors suggest that the development of different cortical areas is the result of parallel input pathways. The auditory cortex is not viewed as simply a collection of various cortical subdivisions with individual functions. Nor does the function of a specific subdivision simply reflect the function of a single pathway. The arrangement of the belt cortices around a center core area and the complex overlaps in inputs show the auditory cortex to be an integrated system. This evidence to support parallel input, distributed processing and integration of function is stated to be the main contribution of the study.
Pontes et al. (1975) conducted a study of the auditory corticothalamic fibers projecting to the MGB in the cat. The study made use of both fiber degeneration and autoradiography. Lesions were made in each auditory cortical subdivision (AI, AII, Ep, SF, Ea, INS). Injections of [3H]leucine were made in AI, AII and SF. The use of complementary methods made up for the shortcomings of each method. The use of silver impregnation with lesioning gave very precise visualization of the organization of corticothalamic fibers but was not as clear as [3H]leucine in the labeling of possible terminals en passage. The use of radiography also overcame uncertainties involving damage to fibers en passage when subcortical white matter is disturbed in lesioning. Problems with the failure of [3H]leucine to be taken up by certain populations of neurons may have resulted in the failure to label all pathways.
The results show fibers projecting from AI passing through the deep dorsal and magnocellular divisions of the MGB and terminating in PL. AII was found to send fibers to the entire dorsal division and to the caudal pole of the MGB. Lesions to Ep were reported to produce degeneration in all areas of the MGB except ZM. The suprasylvian fringe area was the only peri-auditory area to project to areas outside the MGm. Degeneration from SF lesions was found in DD, DS, SG, and several nuclei in the posterior thalamus. Both SF and Ea projected to the magnocellular division. The insular cortex was not found to project to the MGB, however it did project heavily to lower auditory nuclei.
The findings present areas AI, AII and Ep to be reciprocally connected with their respective nuclei in the MGB. The peri-auditory areas SF, Ea and INS were not found to maintain reciprocal connections with their areas of thalamic input. The study is noteworthy for this demonstration of differential connectivity of core and belt areas with the MGB in terms of the principle of reciprocity.
Andersen et al. (1980) is a key paper in the literature on auditory thalamocortical connectivity. It is particularly significant in the present review because it explicitly address the issue of reciprocity in the connections between auditory cortex and thalamus. Its results are interpreted as being supportive of the idea of reciprocity being a general principle in the sensory thalamocortical connectivity. The authors stress the importance of research into the functional role of reciprocal connections in the auditory forebrain.
The study involved microelectrode best-frequency mapping of cortical areas AI , AII and AAF and the use of anterograde (4,5 3H-l-leucine)and retrograde (HRP) tracing to establish the pattern of afferent and efferent connections between these cortical areas and specific subdivisions of the auditory thalamus. The specific findings of the organization of reciprocal thalamocortical connectivity are as follows. The pattern of projections between the MGB and the cortical areas described are (with minor qualifications) entirely reciprocal. Thus the pattern of corticothalamic projections can be inferred directly from the description of the thalamocortical connections. The primary auditory field and the anterior auditory field receive inputs from all subdivisions of the MGv, and from dd, MGm and Pol. They differ in that the cells in MGv projecting to AI form a continuous sheet across that division, while those to AAF are discontinuously arrayed and less dense. The connections are topographically organized and believed to be cochleotopically arranged. This description of the connectivity of AAF is the first reported in the literature; the field was first reported by Merzenich et al. (1975). AII receives its projections from the MGm, Dc and VL. Thus the two tonotopic fields, AAF and AI, reportedly receive similar inputs from the same thalamic divisions while the nontonotopically arranged AII is the target of different thalamic nuclei.
The finding of similarity between the connections of the tonotopic cortical fields AI and AAF and their related thalamic nuclei, compared to the different set of connections between AII and the MGB, led Andersen et al. to offer their results as evidence for two separate and parallel projection systems. These were discussed as being complementary to the essential and sustaining projections of Rose and Woosley (1949; see above) and to the lemniscal and lemniscal-adjunct lines of Graybiel (1973). The later work of Morel and Imig (1987) produced different overall results in terms of connections and called into question this interpretation.
Another significant finding that Andersen et al. reported is the pattern of connections between (presumably functional) columns found in the MGB and cortex. The authors reported differing results in labeling depending on the size of the injection. Large injections of anterograde tracer in the higher frequency representation areas of AI and AAF resulted in a continuous pattern of labeling, while small injections produced a periodic and discontinuous pattern. This three dimensional array of parallel columns was interpreted as indicating that there were alternating bands in the thalamus and tonotopic cortex that matched up with each other. They suggest that reciprocal connectivity creates a pattern of convergent/divergent projections. A point in the cortex receives converging inputs from parallel bands in the MGv and sends divergent projections back these same bands to create the reciprocal connectivity described above (see figure 8a).
Figure 8. a, Schematic representation of the "divergence/convergence"
architecture in the isofrequency dimension, as suggested for the cat. The
lines between the thalamic and cortical IFCs represent the anatomical connections
between both structures; for clarity only the EI-part is shown. b,
Schematic representation of the "point-to –point" architecture, as described
in guinea pig.
(Figure and caption taken from Brandner and Redies, 1990)
An investigation by Redies et al. (1989) into the auditory thalamocortical organization of the guinea pig produced a differing pattern of connectivity than that reported by Andersen et al. (1980) in the cat. Rather than a convergent/divergent pattern as described above, their study describes a "point-to-point" architecture (see figure 8b). To resolve the differing results and confirm the differences between the two species they conducted a study in the cat (Brandner and Redies,1990) as a follow up to the study by Andersen et al. Brandner and Redies (1990) made use of multiple retrograde tracers (HRP, nuclear yellow and Bisbenzimid) simultaneously in the same animal to provide a clearer picture of the actual connectivity than that given by Andersen et al. who had compared injections across animals to obtain their results. Brandner and Redies concluded that the cat had the same pattern of connectivity as seen in the guinea pig.
Morel and Imig (1987) examined the connectivity from the MGB to tonotopically organized cortically fields. Their scheme of electrophysiologically defined organization of auditory cortex is presented in section 2.3 and their division of the auditory thalamus is described in section 2.2.1. An earlier report, Imig and Morel (1983), used a less detailed division of the MGB and did not distinguish between "tonotopic" and "nontonotopic" thalamic nuclei. The results reported in the 1983 paper are summarized in table 2 for comparison to those discussed below.
Table 2 Thalamic nuclei projecting onto auditory cortical fields
(recreated table taken from Imig and Morel, 1983)
In Morel and Imig (1987), the tonotopic fields A, AI, AII, P and VP were examined using a combination of microelectrode mapping and retrograde neural tracing. HRP was used alone or combined with either tritiated bovine serum albumin or nuclear yellow. Injections were made into sites in one or two cortical fields in locations identified after mapping their best-frequency organization. The middle frequency representation, defined as 10-kHz, was often used as a the standard site in order to allow for a number of comparable cases. The frequency selectivity of the auditory thalamus was also mapped. The results of the study show the pattern of connectivity between the MG and functionally determined subdivisions of the auditory cortex, and additionally, reveal the connectional organization within cortical and thalamic divisions based on best-frequency responses.
A summary of specific findings is as follows. Each tonotopically organized cortical area receives inputs from at least one "tonotopic" thalamic area (V, Po and d) and at least one "nontonotopic" thalamic area (Mg, Ms, cd and vl). Inputs are classified as major or minor; each cortical area receives a combination of major and minor inputs. Primary auditory cortex receives major inputs from V, Po and Mg and minor inputs from vl and d. The anterior auditory field has major inputs from Po and Mg and a minor projection from V. The posterior auditory field receives its major inputs from V, dl and d and minor inputs from Po, Mg, Ms and cd. Lastly, the ventroposterior field has inputs from vl, cd and V and minor inputs from Mg, Ms and d (see figure 9).
In the thalamus, areas V and Po were found to contain discrete subdivisions whose neurons mostly projected to a single cortical area (major projection) but contained a portion of neurons that projected to one or more additional cortical areas (minor projections). Also, a small percentage of neurons projecting to areas of tonotopic cortex were found to send collaterals to other tonotopic areas. Of the minor projection fibers from Po and V, a significant number were collaterals from major projection fibers.
A comparison of the results from Morel and Imig (1987) with those of Winer et al. (1977) reveals a difference in the areas to which MGv is said to project. Winer et al. (1977) clearly state that the MGv projects only to AI, while Morel and Imig (1987) report finding projections to A, V and VP as well as AI. The later authors suggest that these areas were overlooked in the previous study. A direct comparison of results is made difficult by differences in parcellation of the cortex and the corresponding nomenclature.
Figure 9. Summary of the major (heaviest) projections from seven
thalamic nuclei to the four tonotopic fields. Minor projections not shown.
The ventrolateral nucleus (vl) has heavy projections to both fields V and
VP. The caudodorsal area has dense projection to field VP. Both Mg and
Po have heavy projections to fields A and AI, although lighter projections
to the latter. V has heavy projections to fields AI, P, and VP, but the
AI projection is the largest. The dorsal cap nucleus has a heavy projection
to field P and DZ.
(Figure and caption taken from Morel and Imig, 1987)
The reported finding by Morel and Imig (1987) of combined inputs from both tonotopic and nontonotopic areas of thalamus to tonotopic auditory cortex conflicts with earlier studies by Andersen et al. (1980; see above) and of Graybiel (1973). These reports describe largely segregated streams of connectivity between the MGB and auditory cortex, with nontonotopic (tonotopic) thalamus projecting to nontonotopic (tonotopic) cortex. However, the possible role(s) of convergent pathways to tonotopic areas of cortex was not discussed by Morel and Imig.
The above studies detailing the topographic organization of cat thalamocortical and corticothalamic connectivity provide an overview of the development of the literature on the neural connectivity between the cortex and MGB in the cat. This review of the literature reveals the immature state of auditory neuroanatomy as seen in the competing and continually evolving nomenclature and organizational schema. The lack of standard terminology often makes direct comparison of results difficult. The early state of auditory neuroscience is also evident from the minimal correspondence of anatomical description with physiology and behavior, or alternately, the considerable speculative leaps of interpretation that are attached to fairly modest experimental results.
Next are reviewed a study examining the laminar origins of projections from AI onto the MGB (Kelly and Wong, 1981) and two studies exploring the terminal morphology of corticothalamic fibers. Ojima (1994) looks at the differences between the fiber and terminal morphology of cells located in layers V and VI of cat primary auditory cortex, while Bajo et al. (1995) reports on the morphology and spatial distribution of corticothalamic fibers and terminal from the various cortical fields in the cat.
Kelly and Wong (1981) made injections of HRP into the MGv of the cat. Cells were labeled throughout layer VI and in the upper portion of layer V. The two laminae represent distinct cell populations and were clearly separated by lower layer V which was almost entirely free of labeling. Approximately 50% of the cells in layer VI were labeled, suggesting that the percentage of cells in this layer projecting to the MGv is likely even higher.
Ojima (1994) examined the distribution and terminal morphology of corticothalamic fibers projecting onto the MGB and IC from layers V and VI of the primary auditory cortex in the cat. Anterograde tracers (PHA-L or biocytin) were injected into single loci that included both layers V and VI. Fibers were found to project to the MGB, Po and IC and ended in two distinct types of terminal endings with differential targets. The projections were primarily distinguished by size and categorized as small or large. Small fibers (~0.4 m m) had small boutons (~1 m m) and were concentrated in the MGv and MGm. Large fibers (~0.8) had larger boutons (~2 m m) arranged in clusters and were found labeled in the MGd, VL and some in the MGm.
Injections were then made that were restricted to either layer V or VI to test for a differential pattern of results. The injections to layer V were found to produce labeling of the large boutons in the MGB, while layer VI was the source of fibers with small boutons. Given the distinct origins, targets and morphology of these two top-down pathways (see figure 10), the authors propose that they must have different functional roles. The layer VI neurons are suggested to provide gain control or a gating function in the MGv while the layer V neurons are speculated to have a more general, modulatory role. These respectively proposed functions do not seem to mesh with their further suggestion that the large cluster-like boutons of layer V neurons might provide a more secure synaptic transmission that that provided by the small boutons of layer VI.
Bajo et al. (1995) also examined the corticothalamic terminals from cat auditory cortex. They looked at the spatial distribution of efferent terminals in the MGB from different auditory cortical fields. Biocytin injections were made into electrophysiologically identified sites in AI,AAF,PAF and AII and resulting patterns of connectivity with the thalamus were recorded and are summarized in figure 11. The terminal boutons were noted to be of large finger-like (5-10 m m) and small spherical (1-2 m m) varieties.
Figure 10. Schematic outline of the dual feedback systems originating
from the cat primary cortex and the intracortical collateralization of
the cells of origin. These systems are completely segregated from each
other and differ in the intrinsic organization of horizontal collaterals,
the size and geometry of their terminal boutons, and their distribution
in the MG and IC.
(Figure and caption taken from Ojima, 1994)
The findings for AI and AAF were reported as being very similar. The rostral portion of the LV was the main target of fibers from AI and AAF. The terminal fields in LV were reported as clearly defined curved strips and were made up of the small spherical boutons. Additionally, smaller, lighter fields of the small spherical terminals were seen in the MGm and RE, while both large and small terminals were found in the MGd and POl.
The PAF was found to project to fields of small terminals, mainly in the caudal MGv and but also in the MGm and RE. An elongated band of large terminals was labeled near the dorsal surface of the MGd.
The main target of AII was the dorsal nucleus of the MGd with primarily small terminal endings along with a few large terminals. Secondly, a dense but restricted field of mixed large and small terminals was found in the deep dorsal nucleus. Finally, a slight projection of small terminals was labeled in the MGm.
The authors interpret their results as supporting a schema of twin reciprocal thalamocortical and corticothalamic pathways. The small terminals are found in areas that are the main source of ascending input to the cortical fields from which the descending fibers arise. Thus the tonotopic cortical areas (AI, AAF and PAF) project to the tonotopically corresponding areas in MGv from which they receive their inputs while nontonotopic AII mainly terminates in the dorsal nucleus of MGd. These small terminal feedback projections are variously speculated to have functional roles including feature detection, gain control, attention or signal optimization. The large terminals are mainly found where tonotopic (nontonotopic) cortical fields project to a major relay MG nucleus of the nontonotopic (tonotopic) pathway. The large boutons are suggested to allow for highly effective synaptic transmission and provide for interaction between the segregated informational streams of the tonotopic and nontonotopic pathways.
A comparison of figures 10 and 11 show considerable overlap in the description of the corticothalamic fibers rising from AI. They also show some differences such as the finding of large terminals in MGm by Ojima (1994) but not by Bajo et al. (1995). Furthermore, the general results and methods of Ojima and of Bajo et al. raise several questions. First, do the large and small terminals described by the two studies belong to the same neural populations and if so why the considerable discrepancy in size of the large terminals? Secondly, it would appear that the Bajo et al. was unaware of the work by Ojima; given the results showing the pattern of differential projections based on laminar origin, how does the lack of control for laminar placement of biocytin by Bajo et al. confuse the outcome of their results?
Figure 11. Schematic representation of organization of corticothalamic
projections. The large finger-like terminals and small terminals are shown
as rising from common axons for simplicity of illustration. The table provides
a qualitative proportion of large and small endings in the various thalamic
nuclei from projections from specific cortical fields.
(Figure taken from Bajo, 1995)
Ryugo and Killackey (1974) lesioned the MGv and MGm in the rat to examine their patterns of thalamocortical projections. The results showed that these two divisions could be separated by differential projections revealed by fiber degeneration. Large unilateral lesions of the MGB resulted in dense terminal degeneration in a portion of the caudoputamen though most degenerating fibers projected to the neocortex. The heaviest degeneration in the cortex was seen in layers III and IV, with lighter fiber and terminal degeneration seen out to layer I. Discrete lesions were then placed in the ventral and medial divisions to attempt to separate the overall pattern into component contributions.
Anterograde degeneration from lesions restricted to within the MGv was seen in "large caliber fibers" which project to a restricted focus in neocortex and terminate in layers III and IV. Lesions restricted to MGm resulted in degeneration of "small caliber fibers" which gave off collaterals to the caudoputamen and terminated in broadly in all cortical layers including layers I and II. The study has several weaknesses: the report did not quantify ‘small’ versus ‘large’ fibers; there was no specification of what specific areas of neocortex contained degenerating terminals; and there was no reported attempt to verify the nuclear boundaries in the MGB where the lesions were placed. Nevertheless, the study provides several significant results. The findings of a diffuse pattern of termination from the MGm and its sending collaterals to the caudate putamen provide an explanation for the failure of Rose and Woosley (1949) to find degeneration in the MGm following cortical lesion. Ryugo and Killackey suggest that the collateral fibers to the caudoputamen and the wide distribution of terminals in the cortex serve to "sustain " the cells of the MGm when the cortex is lesioned. More importantly, the demonstration of distinct patterns of thalamocortical connectivity in the rat supports the idea that parallel processing of sensory information is a basic principle of thalamocortical organization in the mammalian brain.
Patterson (1977) provided a description of the projections from the MGB to the auditory cortex in rat based on both anterograde degeneration and retrograde axonal transport of HRP. Table 3 provides a summary of his findings of auditory thalamocortical connectivity. The ventral division of the MGB was found to project to the what Patterson called the core cortex which corresponds to primary auditory cortex. Based on the lack of labeling of cells in MGv from HRP applications in the belt cortex, Patterson tentatively concluded that the MGv projects only to the core cortex. Later studies, discussed below (Arnault and Roger, 1990; Romanski and LeDoux, 1993b), provide evidence that the ventral MGB does send projections to secondary auditory cortex. The MGv has a pattern of topographic connectivity with the core cortex in both the antero-posterior and dorso-ventral dimensions. Lesions placed along an antero-posterior continuum in the MGv resulted in a corresponding pattern of degeneration in the core cortex along an antero-posterior axis. In the dorso-ventral dimension, there is an inverse mapping, with dorsal MGv projecting to the ventral core cortex and ventral MGv sending projections to dorsal core cortex. The comparison of this result with findings from other studies will be made below in the review of Romanski and LeDoux (1993b).
The shell nuclei of the auditory thalamus, consisting of the dorsal division, caudal division, marginal zone and ventrolateral nucleus, project to the belt cortex. The belt cortex forms a crescent shaped ring lying posterior to the core cortex and partially encircling it. Each of the shell nuclei was found to project to a specific portion of the belt cortex. The ZM was found to project to the dorsal belt cortex. The caudal division projects to the antero-ventral area of the belt cortex. Both the dorsal and ventrolateral divisions of the MGB were found to have a topographic mapping onto cortex. The dorsal division projects to belt cortex exhibiting both medial-lateral and antero-posterior axes. Posterior MGd projects to ventral belt cortex while anterior MGd projects to dorsal belt cortex. Lateral MGd projects to a crescent shaped area adjacent to the core cortex while medial MGd terminates in a crescent outside of this inner crescent. In the case of the ventrolateral nucleus, anterior VL projects to postero-ventral belt cortex while posterior VL terminates in the antero-ventral portion of belt cortex.
The laminar terminations of both MGv and the shell nuclei were found to be concentrated in cortical layers III and IV. Slight degeneration of terminals was sometimes also seen in layers I and II.
The pattern of projection of the medial division of the MGB was described
as diffuse. Large areas of auditory cortex demonstrated degeneration following
lesions in MGm. These areas included much of the core cortex as well as
cortex lying anterior and dorsal to it. The belt cortex was not found to
receive significant projections from MGm either by retrograde labeling
or anterograde degeneration. The laminar termination of fibers from MGm
were found to be distributed throughout the cortical laminae.
|
Cortical Target
|
|
|
|
|
Belt Cortex
|
|
|
|
|
Core Cortex
|
|
|
|
Vaughan (1983) examined the organizational pattern of extrinsic inputs to the rat auditory cortex from the MGB and the contralateral cortex. The Fink-Heimer method was used to visualize degenerating terminals in the cortex resulting from lesioning the MGB and/or cutting the corpus callosum. Details of cross callosum connectivity will not be covered in this literature review but it is important to note that a full understanding of cortical connectivity requires attention to the interaction of a variety of input patterns.
The study did not provide a description of the topographic projections from the MGB to the auditory cortex. Electrolytic lesions to the MGB were large, and intended to destroy that entire structure. This resulted in widespread degeneration in the areas of the auditory cortex designated 41, 20 and 36 by Krieg (1946a; see figures 12a,e). The degenerating terminals were found concentrated in a horizontal band consisting of cortical layer IV and deep layer III extending over both primary and secondary cortex. Additionally, a lighter band of terminals was seen in layer VI. The degeneration in layers III, IV, and VI were seen as forming alternating patches of heavy and light terminations. A band of terminals was also seen in the most superficial part of layer I which was evenly distributed in its density.
Callosal inputs were seen to form a trilaminar distribution. Fibers terminated in 1) layer I through III, excluding outer layer I; 2) outer layer V and ; 3) in layer VI. While the callosal bands in area 41 are continuous, there is a variation in their density in anterior and dorsal area 41 where they form alternating patches of heavy and light terminations. The remainder of area 41 and areas 36 and 20 are uniform in their callosal terminal distributions.
The overlap of the two systems of extrinsic inputs was demonstrated by double lesioning. In the laminar dimension, the combined systems provide inputs to all layers of cortical columns except lower layer V. The parts of area 41 that showed alternating patches in both thalamic and callosal terminations showed an evenly distributed terminal distribution with the combined lesion. This is presumed to be the result of the interweaving of alternating patches of callosal and thalamic inputs. The remainder of rat auditory cortex was seen as having converging overlap of callosal and thalamic inputs.
The study is significant for pointing out the complex three-dimensional integration of separate systems of extrinsic inputs to auditory cortex as illustrated by the interaction of segregation and overlap in both laminar and areal organization.
Winer and Larue (1987) performed a study of the pattern of reciprocal connectivity between the auditory cortex and MGB in the rat. A mixture of HRP and [3H]leucine was injected into various locations in auditory cortex and the resulting anterograde labeling of terminals and retrograde labeling of somata were compared in the MGB by examining adjacent sections alternately processed for each tracer. The authors report that their main finding is that while there was a general overlap of HRP labeled cell bodies and audiographic silver grains in corresponding areas of adjacent sections, there were many areas that lacked reciprocal connections. These zones of non-reciprocity were found in all divisions of the MGB. The MGv received relatively heavy corticothalamic projections for the number of cells projecting to auditory cortex while the MGm showed only sparse radiographic labeling for the number of HRP filled cells. The dorsal division was said to be essentially congruent in the balance of cell and terminal labeling. I would have to guess that the criteria for what was equivalent labeling of cells and terminals was intuitive. The authors did present a quantitative methodology for comparing the density of labeled terminals in one zone (14,000 m m2) with those in another with assignment of a level from I-IV depending on the count of silver grains.
The authors point out several considerations relating to their methodology. The results were based on the comparison of adjacent sections rather than allowing for the unambiguous observation of cells and terminals labeled in the same section. This loss was viewed as offset by the improved sensitivity afforded by separate processing of the two tracers used. This sensitivity was viewed as crucial in answering the question of whether global patterns of reciprocity and non-reciprocity existed in the MGB. They also considered the potential problem of whether the two methods used were equally sensitive in diffusion, uptake and transport. They dismissed this concern by noting that every division of the MGB received some labeling and also that the results of the audiographic labeling were confirmed by matching results provided by (unintended) anterograde labeling by HRP. The consideration of differential shrinkage by the different methods of processing was adequately taken into account by several strategies.
The following limitations and observations, not discussed in the study, are also noted. The tracer injections were large and not restricted in their spread to single subdivisions of auditory cortex. Consequently, the study did not provide detailed information either on the areal pattern of connectivity between the cortex and MGB or on laminar patterns of projection and termination. Also, without punctuate injections, a precise pattern of connectivity is not established. What is demonstrated is that a given point sends projections to some point in a large area of cortex and receives terminals from some point, possibly the same or possibly a distant point, in the same large area of cortex. Highly restricted focal injections would better demonstrate if there is a point-to-point reciprocal connectivity or a some other pattern.
The study is nonetheless unique in the literature as an explicit attempt to map the extent of connectional reciprocity between the MGB and auditory cortex in the rat. It represents a starting point in mapping a detailed picture of thalamocortical reciprocity including the issues of laminar and areal relationships. It also emphasizes the widespread extent of top-down projections in the upper auditory system and raises the question of their functional significance of the existence of reciprocal versus non-reciprocal areas in the MGB.
Figure 12 A-C shows parcellation schemes of rat auditory cortex
as labeled.
(Figure taken from Scheel, 1988)
Figure 13 A Injection centers in fields A-D of core area are
represented by their experimental numbers. B shows the location
of labeled neurons in the segments of MGv as a result of injections demonstrated
in A.
(Figure taken from Scheel, 1988)
Scheel, 1988 made use of iontophoretic injections (1-5 nl) of WGA-HRP into auditory cortex to detail a schema of topographical organization of the auditory cortex and the ventral division of the MGB based on topography of thalamocortical connectivity. Scheel claims to have made several original findings. The cortical core area and MGv were divided into four parts each. The core cortical auditory area is divided by borderlines running dorsoventrally. Seen from a coronal section, the divisions of the MGv are wedge shaped sectors giving the impression of a pie cut into rough quarters. Beginning with the medial sector of the MGv being mapped topographically onto the rostral most fields of the core area, moving counterclockwise, the sectors are mapped onto the core area from rostral to caudal. This organization, along with the numbered injection sites form a composite of the results from all the animals in the study, and is illustrated in figure 13. The use of small injections resulted in the labeling of small restricted areas in the MGv, no labeling patterns of laminae or bands were found.
The results of the study are at face value quite remarkable. However the composite results across animals resulting in the parcellation of auditory cortex as seen in figure 12 and the subdivision of the core area and the MGv shown in figure 13 have an implied degree of precision and accuracy that cause me to doubt their veracity. The experiment would certainly be worth repeating to attempt to replicate the results.
Roger and Arnault (1989) examined the connectivity between the MGv and primary auditory cortex in the rat with an emphasis on the topographical pattern of thalamocortical projections. Discrete applications of crystalline WGA-HRP were made to limit diffusion. The WGA-HRP was taken up and carried by both anterograde and retrograde transport. The areas of retrograde and anterograde labeling in MGv were coincident, demonstrating reciprocal connectivity between Te1 and the MGv. In addition to projections from the MGv, Te1 received inputs from the MGd, primarily the deep dorsal nucleus. There was also labeling in Pol and the MGm.
The results were said to suggest a topographical order of projection with a mediolateral shift in the MGv producing a rostrodorsal to caudoventral progression across Te1, and rostrocaudal shift in the MGv resulting in a dorso-ventral progression in Te1. The MGd had a similar topographical organization except that there was only limited connection with the ventral part of Te1. The MGm was not found to have a discernable pattern of topographic organization. A comparison of these results with those from other studies is found below with the review of Romanski and LeDoux (1993b).
Arnault and Roger (1990) examined the connections of the secondary auditory cortices in the rat by means of anterograde and retrograde transport of WGA-HRP. The results of injections to Te2 and Te3 were similar but each did possess unique features. The thalamocortical connectivity of the secondary cortices were quite distinctive from that of the primary auditory cortex reported above. Te2 was not found to be connected to MGv; the greatest labeling was found in the ventrolateral MGd and the lateral peripeduncular area. The caudal MGd and the caudodorsal nucleus were also labeled along with slight labeling in the MGm and SG. Te3 was likewise mainly connected with the MGd and PPA, but with the medial portions of those structures, with significant labeling also seen in the caudal MGm. Connectivity with Te3 was also revealed in light labeling in the SG, MGcd and MGv. The authors reported that there was no indication of any topographic organization between the MGB and Te2/Te3.
The study also reported on extrageniculate afferent and efferent projections in the thalamus, brainstem and contralateral homotypic cortex, however no mention of corticocortical connectivity was made.
Rouiller et al. (1991) made injections of PHA-L into rat primary auditory cortex and examined the corticothalamic projections shown by anterograde labeling of axons and terminals. Corticothalamic fibers gave rise to two types of terminal endings: small boutons (~1 m m) and "giant’ terminals (5-10 m m). The small terminals were found in the MGv, Pol, MGd and RE. The larger terminals were found only within the ventral part of the MGd. The finding of two types of terminal morphology with distinct patterns of corticothalamic targets suggest the existence of two distinct functional mechanisms. These functional roles were discussed above in the review of the latter, similar study by Bajo et al. (1995).
Figure 14 Lateral view of the brain depicting the auditory cortical
fields. The primary auditory region (TE1) is depicted in relationship to
surrounding auditory cortical belt regions TE2c, TE2d, TE3v and TE3r (based
on Zilles et al. 1990)
(Figure and caption taken from Romanski and LeDoux, 1993b)
Romanski and LeDoux (1993b) performed a study of the pattern of thalamocortical projection from MGv in the rat. Previous studies in the rat had never made use of anterograde tracer for this purpose. The use of anterograde tracer phaseolus vulgaris leucoagglutinin (PHA-L) was deemed to be superior to retrograde methods to demonstrate the laminar and areal patterns of termination of fibers from the MGv. (The areal organization of auditory cortical fields is shown in figure 14.) The results showed the MGv projecting primarily to the middle layers of Te1. Injections placed in the OV division of the medial MGv densely labeled Te1 while largely sparing Te1v and with no labeling of secondary auditory cortex. Placed more laterally, injections including LV also revealed fibers heavily projecting to Te1 and lightly to Te1v, but additionally showed projections to areas of Te2 and Te3. A direct rostral-caudal mapping was found with caudal MGv projecting to caudal Te1 and Te2c while rostral MGv sent projections to rostral Te1 and Te3r. This is in agreement with Patterson (1977), Winer and Larue (1987) and Scheel (1988) but in contradiction with the results of Roger and Arnault (1989).
Romanski and LeDoux propose that the projections from the specific auditory thalamic nucleus, the MGv, overestimates the areal boundaries of primary auditory cortex and conclude that primary auditory cortex (Te1 excluding Te1v) can be delimited as the cortical area that receives fibers from both the OV and LV divisions of the MGv. Figure 14 shows the arrangement of the auditory cortical fields as seen from a lateral view of the brain.
The authors make a call for the development of a comprehensive view of rat auditory cortex through an integrated approach to research including physiological, cytoarchitectonic, connectional and neurochemical techniques.
2.4.2 Corticocortical connections
The literature on auditory corticocortical connectivity in the rat is virtually non-existent. Reports by Romanski and LeDoux (1993a) and Mascagni et al. (1993) touch on the subject but are do not address issues of laminar organization, topographic organization or reciprocal connectivity. Clancy (1996) examined the reciprocal connectivity between the primary and secondary auditory cortices in the rat.
The use of anterograde (DA-TR) and retrograde (DY) fluorescent tracers were reported to demonstrate a pattern of reciprocal connectivity between primary and secondary cortex. Injections of DA-TR were made in Te1 and ipsilateral projections found in Te2 and Te3. No details of laminar origins or termination were reported, nor were any patterns of topographical organization. Surface applications of DY were placed in Te1 and retrogradely cell bodies were found labeled in Te2 and Te3. The cells were reported to have been labeled in both the deep and superficial layers. Layer VII (deep layer VI) cells were shown to form a distinctly labeled lamina. However, the specific contributions of various laminar origins of top-down projections was not explored. No control was provided to ensure that the labeling observed in Te2 and Te3 were specifically from layer I. The appearance of labeled cells in the middle layers of Te1 below the site of the surface application allows for the possibility that deeper penetration of the tracer beyond layer I may have labeled additional cells in the secondary cortices. Other than generally stating that the ipsilateral connections between Te1 and the secondary cortices were reciprocal, no descriptive details of the pattern of reciprocity were given and no figures were provided that illustrated the relationship between the anterograde and retrograde labeling.
Injections of DA-TR were made in the secondary cortices (Te2 and Te3)
and the results labeled projections to Te1. The top-down projections were
said to avoid layer IV and terminate primarily in layer V and in layer
I of Te1. The fibers projecting to layer I terminated in long horizontal
fibers parallel to the surface. The pattern of long horizontal fibers was
clearly illustrated in their photomicrographs. The top-down avoidance of
layer IV was not so clearly illustrated in camera lucida drawings intended
to show the preferential labeling of layer I.
3. Interactive Auditory Processing, Reentry, and Behavior
Reciprocal neural connectivity creates complex patterns of interaction between and within the hierarchical levels of the auditory system: thalamocortical loops; corticocortical loops; and parallel segregated, afferent pathways diverging and converging to form multiple areas each with its unique pattern of inputs and outputs. How does this anatomy give rise to auditory related behavior?
As previously noted, there are two types of interactivity of interest in the paradigm of interactive sensory processing, the active engagement of an organism with its environment through the interaction of the sensory and motor systems and the interactivity between bottom-up and top-down neural signaling. How are these related? Clearly they possess an overlap with each other in that the traditional view of sensory processing and behavior describes the afferent input of sensory information causing efferent motor output in response. The modification of the traditional view provided by the interactive paradigm is the emphasis that motor output is not simply the final stage of cortical processing. Motor control and sensory perception are part of a continuous output/input (input/output) loop that has motor control selecting and actively probing the sources of sensory input which in turn provide the information to select and guide the next movement. However, as we have seen in the above review of literature on the anatomy of the auditory forebrain, there are extensive efferent sensory projections in the auditory system. Consequently, this leads to the question of what role, if any, top-down neural connectivity plays in auditory perception. In the introduction to this paper, I stated that the interactive paradigm of sensory processing provides a framework to study how phenomena such active listening and selective auditory discrimination take place in the brain. I will now examine how top-down sensory connectivity might provide the neural basis for these perceptual abilities.
According to Churchland et al. (1994), "interactive vision is exploratory and predictive". Is hearing interactive? If so in what ways? The head and the eyes move to explore the world resulting in the selection of input to be processed. Ears can be directional, as with dogs, or the head can be turned to give better reception, but in general, the act of hearing in humans is passive from the point of view of motor control. Nevertheless, our ability to hear is very selective. One can attend to a specific auditory source in the presence of competing background noise, suppressing unwanted surrounding stimuli. At the same time, background sources are seemingly monitored for high interest stimuli, such as the listeners own name, i.e. the cocktail party phenomena (Cherry, 1953) The ability to shift selective attention between sources in the auditory environment corresponds to the exploration of the visual world through eye movement. The fact that listening is a behavior that is under conscious control in humans, implies that it is a neocortical process. Top-down neural connectivity in the auditory system may provide the neural basis for active listening through the mechanism of selective attention that is derives from proposed top-down functional features such as gain control (Winer, 1992; Ojima, 1994), filtering, and selective signal amplification/suppression (Winer, 1992).
Perceptual invariance is an example of predictive aspect of interactive hearing. This feature of our hearing system can be seen as assisting selective auditory discrimination by reducing the number of possible choices to discriminate between and thus providing a solution the problem of detecting behaviorally significant signals from competing inputs, including noise.
The ability to categorize phonemes as invariant units is learned early in life. At birth infants can distinguish sounds used in language across a continuum that includes the phonetic categories of all human languages By 6 months of age their perceptual phonemic boundaries have began to develop to the language(s) to which they have been exposed (Eimas et al., 1971; Kuhl, 1990). The combination of phonetic perceptual invariance and the order imposed on language by grammatical rules allows the brain to process auditory information as language quickly and efficiently by limiting the number of possible assignments. In the case of the raw auditory stream, the number of phonemes that can be assigned is limited. The assignment of phonemes to words is restrained by the available lexicon of the language. In the case of semantic meaning, the order of the words may restrict the combinatorial explosion of meanings that would result from the multiple possible assignments available to each word (Charniak, 1993).
All of these examples of limitations on the numbers of possibilities
of phonemic category, word identity, and semantic meaning of language,
are examples of predictive auditory processing. It is predictive in that
the auditory-language system has preexisting sound and language templates
that it expects to match. This allows for the accommodation of speech production
errors and the resolution of lexically ambiguous language. Top-down neural
connectivity in the auditory system may provide the neural basis for selective
auditory discrimination as seen in relation to predictive functions to
facilitate language. This might derive from top-down projections that allow
for perceptual invariance by pattern matching, categorization, priming
and the comparison of predicted input with actual sensory input (Cauller,
1995)
4. Conclusions and future directions
As illustrated in this literature review, knowledge of the anatomical and functional properties of the auditory forebrain has advanced greatly in the last 50 years. There is an increasing recognition of the existence and magnitude of reentrant and top-down connectivity and of the importance of accounting for these features when attempting to create neurobiologically realistic explanations of auditory processing and behavior. However, little effort or progress has been made in actually combining these new anatomical and functional studies with the purpose of exploring the neural correlates of behavior. Neuroscience is inherently multidisciplinary, but there is a need for greater interdisciplinarity in specific research designs to resolve the complex interaction of the various levels of analysis spanning the spectrum from the molecular at one end to behaving systems at the other.
The combined works of Kulics (1977, 1982) Cauller and Kulics (1991), Cauller (1995), Pandya (1995) and Clancy (1996) suggest an interesting line of research. Cauller and Kulics’ finding that backward cortical projections signal conscious touch sensation might be tested in the auditory system to determine if this functional anatomy is modality specific or may be common to other sensory systems. Clancy demonstrated that the pattern of top-down connectivity focusing on layer I, documented by Pandya in monkey auditory cortex, is valid in both rat somatosensory and auditory cortices. The successful demonstration of an evoked-potential component derived from top-down projections onto primary auditory cortex in the rat that predicts conscious auditory discrimination would allow for the generalization of Cauller and Kulics’ findings by extension across both species and modality. A plan of experimental research that could provide a test for the role of top-down convergence on layer I in rat auditory in a behaviorally relevant discrimination task is outlined below. In the following section the anatomical terminology of Paxinos and Watson (1998) will be used.
First, the corticocortical organization of rat auditory cortex would established by extending the anatomical tracing work of Clancy (1996). There are numerous interesting questions regarding reciprocal connectivity between the auditory cortices still unanswered. The larger question, "what is the pattern of reciprocity between the primary and secondary cortices?" can be broken down into the following experiments. Injection of retrograde tracers into the secondary cortices could reveal what cortical layer(s) in Au1 gives rise to fibers projecting to AuV and AuD. Application of anterograde tracer Dextranamine in Au1 should demonstrate the specific layer(s) in AuD and AuV that are the target(s) for these projections. The use of restricted focal injections of anterograde tracer in primary cortex could test if fibers terminate in a single, similarly focal area in secondary cortex (i.e. is there a point-to-point connectivity or some other organization?). Similarly, a careful orderly mapping of Au1 using highly focal injections of multiple anterograde tracers might reveal what a topographic organization of Au1 as it relates to projections to the two (or more) identified areas of secondary auditory cortex.
Questions detailing top-down connectivity include the following. From what specific layers in the secondary cortices do projections to Au1 arise; is there a single projection system or might fibers rising from different layers represent distinct functional systems? The use of layer specific applications of retrograde tracer into Au1 might reveal a differential pattern of origins and terminations. If the top-down pattern of projection from the higher order auditory cortical areas to Au1 is a diffuse termination of horizontal fibers in layer I, then is this area of termination centered over the columns that project to the columns in the secondary cortices from which the area of layer I termination originates? This question can be examined two ways using double labeling studies. Focal, combined injections of anterograde and retrograde tracers could be placed in secondary cortex in the layers found to project to Au1. Evaluation of the results would focus on the correspondence of the positions of the anterogradely labeled fiber plexus in layer I and the retrogradely labeled cell columns. Additionally, the results could be verified by making an injection of anterograde tracer in the cell layers in Au1 that project to secondary cortex and an application of retrograde tracer directly above the injection on the pial surface. The degree of correspondence of tracers in Au1 would be compared with the degree of convergence of the two tracers in secondary cortex as seen in the precision of double labeling. Most of the above designs call for the use small, highly restricted injections and applications. The need for highly focal injections is in any case necessary when working in the auditory cortex of the rat. This was pointed out by many of the authors covered in this review who were working with rat preparations (see especially Romanski and LeDoux, 1993b and Clancy, 1996
Finally, it would be of interest to examine the connectivity between the individual areas of secondary cortex to determine their level of connectivity and if they exhibit a hierarchical relationship with each other as defined by Felleman and Van Essen (1991).
Next, electrophysiological baseline data would be collected from anesthetized rats. Recordings should be made in primary and secondary auditory cortices to characterize the three dimensional spatial and temporal distribution of the cortical response to a variety of auditory stimuli. This characterization could assume the form of depth profiles of auditory evoked potentials (AEP) , multiple unit activity (MUA), and current source density analysis (CSD; Cauller and Kulics, 1991, Hamilton, 1997) This data should be significant in providing a basis for interpreting data collected in behaving animals using chronically implanted electrodes. It would also allow a comparison of AEPs for different types of stimuli. Several rats would then be chronically implanted with recording electrodes in primary and secondary auditory cortex. Auditory evoked potentials would then be recorded in a behaving state in response to the collection of auditory stimuli used previously. This data would be compared to the data from the anesthetized animals to determine what features are unique to the behaving state. In particular, it is expected to find components of the rat AEP that correspond to P1 and N1 in the monkey. This is supported by the findings of Barth and Di (1991). These AEP components would need to be tested to determine if they are behaviorally relevant to signal discrimination.
A signal detection procedure would be implemented where rats were trained in an go/no-go auditory discrimination task. They would be trained to respond to a tone stimulus by choosing whether or not to press a bar, indicating whether the tone was "loud" or "soft" in order to get a reward. Data would collected by computer and analyzed for the ratio of correct versus incorrect responses. When a rat reaches a predetermined level of proficiency, it would be chronically implanted with recording electrodes into the primary and secondary cortices. Additionally, a microdialysis delivery system could be chronically implanted to deliver glutamate antagonists to the surface of Au1 in order to manipulate layer I synaptic activity. When the rat recovered from surgery and demonstrated an intact proficiency in the discrimination task, AEPs would then be recorded in response to auditory stimuli while the rat is engaged in the discrimination task. The intensity difference would be reduced to a level where discrimination errors are made. Responses would be grouped as hits, misses, false alarms and correct rejections and the AEPs of each class of response will be averaged. A comparison of these averages would reveal a pattern of relationship between the relative amplitude of neural response for the P1 and N1 components and the corresponding behavioral responses.
The final step would involve the application of glutamate antagonists to layer I of Au1 during performance of the behavioral task. If layer I is responsible for the generation of the identified, behaviorally relevant component (N1), then it should be abolished by the shut down of layer I synaptic activity. Similarly, the ability of the rat to perform the task should be eliminated along with N1. The behavior should return after a period of time as the effect of the antagonist wears off and the neural correlate reappears. Additional reversible pharmacological lesions could be selectively performed in areas of secondary cortices that are identified as the sources of top-down projections to the primary auditory cortex. This might allow for the identification of the specific higher order area(s) that are responsible for the generation of N1 in Au1 (see Jackson, 1997).
This design combines neuroanatomical, electrophysiological and behavioral
techniques in an effort to identify behaviorally relevant AEP components
and their neural origins. It might provide evidence that the behaviorally
relevant N1 found in monkeys is a general feature of mammalian sensory
systems. It might also provide neuropharmacological evidence to support
the top-down origins of behaviorally relevant brain signals such as N1.
Aitkin, L. (1990) The Auditory Cortex: Structural and functional bases of auditory perception, Chapman and Hall, London.
Andersen, R. A., P. L. Knight, and M. M. Merzenich (1980) The thalamocortical and corticothalamic connections of AI, AII, and the anterior auditory field (AAF) in the cat: evidence for two largely segregated systems of connections. Journal of Comparative Neurology 194, 663-701.
Arnault, P., and M. Roger (1990) Ventral temporal cortex in the rat: connections of secondary auditory areas Te2 and Te3. Journal of Comparative Neurology 302, 110-123.
Bajo, V. M., E. M. Rouiller, E. Welker, S. Clarke, A. E. Villa, Y. de Ribaupierre, and F. de Ribaupierre (1995) Morphology and spatial distribution of corticothalamic terminals originating from the cat auditory cortex. Hearing Research 83, 161-174.
Barlow, H. B. (1972) Single units and sensation: a neural doctrine for perceptual psychology? Perception 1, 371-394.
Barth, D. S., and S. Di (1991) The functional anatomy of middle latency auditory evoked potentials. Brain Research 565, 109-115.
Beyerl, B. D. (1978) Afferent projections to the central nucleus of the inferior colliculus in the rat. Brain Research 145, 209-223.
Brandner, S., and H. Redies (1990) The projection from medial geniculate to field AI in cat: organization in the isofrequency dimension. Journal of Neuroscience 10, 50-61.
Brugge, J. F., and R. A. Reale (1984) Auditory Cortex. In Cerebral Cortex Vol. 4, A. Peters, and E. G. Jones, eds., pp. 229-271, Plenum Press, New York.
Burton, H., and E. G. Jones (1976) The posterior thalamic region and its cortical projection in New World and Old World monkeys. Journal of Comparative Neurology 168, 249-301.
Cauller, L. J. (1995) Layer I of primary sensory neocortex: where top-down converges upon bottom-up. Behavioral Brain Research 71(1-2), 163-170.
Cauller, L. J., and A. T. Kulics (1991) The neural basis of the behaviorally relevant N1 component of the somatosensory-evoked potential in SI cortex of awake monkeys: evidence that backward cortical projections signal conscious touch sensation. Experimental Brain Research 84, 607-619.
Charniak, E. (1993) Statistical Language Learning, MIT Press, Cambridge.
Cherry, E. C. (1953) Some experiments on the recognition of speech, with one and two ears. Journal of the Acoustical Society of America 25, 975-979.
Churchland, P. S., V. S. Ramachandran, and T. J. Sejnowski (1994) A Critique of Pure Vision. In Large-Scale Neuronal Theories of the Brain, C. Koch, and J. L. Davis, eds., pp. 23-60, MIT Press, Cambridge.
Clancy, B. (1996) Afferent Convergence in Layer I of Rat Somatosensory Cortex: Neuroanatomical Support for an Interactive Sensory-Motor Paradigm [Dissertation], University of Texas at Dallas.
Clerici, W. J., and J. R. Coleman (1990) Anatomy of the rat medial geniculate body: I. Cytoarchitecture, myeloarchitecture, and neocortical connectivity. Journal of Comparative Neurology 297, 14-31.
Crick, F., and C. Koch (1990) Towards a neurobiological theory of consciousness. Seminars in the Neurosciences 2, 263-275.
Edeline, J. M., and N. M. Weinberger (1992) Associative retuning in the thalamic source of input to the amygdala and auditory cortex: receptive field plasticity in the medial division of the medial geniculate body. Behavioral Neuroscience 106, 81-105.
Edelman, G. M. (1978) Group Selection and Phasic Reentrant Signaling: A Theory of Higher Brain Function. In The Mindful Brain: Cortical Organization and the group Selective Theory of Higher Brain function, pp. 51-100, The MIT Press, Cambridge.
Edelman, G. M. (1987) Neural Darwinism: The Theory of Neuronal Group Selection, Basic Books, New York.
Edelman, G. M. (1989) The Remembered Present: A Biological Theory of Consciousness, Basic Books, New York.
Eimas, P., F. Siqueland, P. Jusezyk, and J. Vigorito (1971) Speech perception in early infancy. Science 171, 303-306.
Faye-Lund, H. (1985) The neocortical projection to the inferior colliculus in the albino rat. Anatomy & Embryology 173, 53-70.
Felleman, D. J., and D. C. Van Essen (1991) Distributed hierarchical processing in the primate cerebral cortex. Cerebral Cortex 1, 1-47.
Friedman, D. P. (1983) Laminar patterns of termination of cortico-cortical afferents in the somatosensory system. Brain Research 273, 147-151.
Galaburda, A. M., and D. N. Pandya (1983) The intrinsic architectonic and connectional organization of the superior temporal region of the rhesus monkey. Journal of Comparative Neurology 221, 169-184.
Gerren, R. A., and N. M. Weinberger (1983) Long term potentiation in the magnocellular medial geniculate nucleus of the anesthetized cat. Brain Research 265, 138-142.
Graybiel, A. M. (1973) The thalamo-cortical projection of the so-called posterior nuclear group: A study with antereograde degeneration methods in the cat. Brain Research 49, 229-244.
Hamilton, P. R. Laminar Analysis of Forepaw-Evoked Potentials in Rat SI Cortex. Unpublished research project, University of Texas - Dallas.
Hashikawa, T., M. Molinari, E. Rausell, and E. G. Jones (1995) Patchy and laminar terminations of medial geniculate axons in monkey auditory cortex. Journal of Comparative Neurology 362, 195-208.
Hill, S. J., and D. L. Oliver (1993) Visualization of neurons filled with biotinylated-lucifer yellow following identification of efferent connectivity with retrograde transport. Journal of Neuroscience Methods 46, 59-68.
Hubel, D. H., and T. N. Wiesal (1962) Receptive fields, binocular interaction and functional architecture in the cat's visual cortex. Journal of Physiology 160, 106-154.
Hubel, D. H., and T. N. Wiesel (1965) Receptive fields and functional architecture in two non-striate visual areas (18 and 19) of the cat. Journal of Neurophysiology 28, 229-289.
Imig, T. J., and A. Morel (1983) Organization of the thalamocortical auditory system in the cat. Annual Review of Neuroscience 6, 95-120.
Imig, T. J., and A. Morel (1985) Tonotopic organization in lateral part of posterior group of thalamic nuclei in the cat. Journal of Neurophysiology 53, 836-851.
Imig, T. J., and A. Morel (1988) Organization of the Cat's Auditory Thalamus. In Auditory Function: Neurological Bases of Hearing, G. M. Edelman, W. E. Gall, and W. M. Cowan, eds., pp. 457-484, The Neurosciences Institute (John Wiley & Sons), New York.
Imig, T. J., and R. A. Reale (1980) Patterns of cortico-cortical connections related to tonotopic maps in cat auditory cortex. Journal of Comparative Neurology 192, 293-332.
Jackson, M. E. (1997) Dynamics of reciprocal corticocortical connections in the somatosensory cortex of the rat: Electrophysiology and computational modeling [Dissertation], University of Texas at Dallas.
Jones, E. G., and H. Burton (1976) Areal differences in the laminar distribution of thalamic afferents in cortical fields of the insular, parietal and temporal regions of primates. Journal of Comparative Neurology 168, 197-247.
Kandel, E. R., J. H. Schwartz, and T. M. Jessel, eds. (1991) Principles of Neural Science, 3rd ed., Appleton & Lange, Norwalk.
Kelly, J. P., and D. Wong (1981) Laminar connections of the cat's auditory cortex. Brain Research 212, 1-15.
Kilgard, M. P., and M. M. Merzenich (1998) Cortical map reorganization enabled by nucleus basalis activity. Science 279, 1714-1718.
Knight, P. L. (1977) Representation of the cochlea within the anterior auditory field (AAF) of the cat. Brain Research 130, 447-467.
Koch, C., and J. L. Davis, eds. (1994) Large-Scale Neuronal Theories of the Brain, MIT Press, Cambridge.
Kolb, B., and I. Q. Whishaw (1996) Fundamentals of Human Neuropsychology, 4th ed., Freeman & Co., New York.
Krieg, W. J. S. (1946a) Connections of the cerebral cortex. I. The albino rat. A. Topography of the cortical areas. Journal of Comparative Neurology 84, 221-275.
Krieg, W. J. S. (1946b) Connections of the cerebral cortex. I. The albino rat. B. Structure of the cortical areas. Journal of Comparative Neurology 84, 277-323.
Krieg, W. J. S. (1947) Connections of the cerebral cortex. I. The albino rat. C. Extrinsic connections. Journal of Comparative Neurology 86, 267-394.
Kudo, M., K. K. Glendenning, S. B. Frost, and R. B. Masterton (1986) Origin of mammalian thalamocortical projections. I. Telencephalic projections of the medial geniculate body in the opossum (Didelphis virginiana). Journal of Comparative Neurology 245, 176-197.
Kuhl, P. (1990) Toward a new theory of speech perception. In Proceedings of the International Conference on Spoken Language Processing Vol. 2, S. Furui, ed., pp. 745-748, Acoustical Society of Japan, Tokyo.
Kulics, A. T. (1982) Cortical neural evoked correlates of somatosensory stimulus detection in the rhesus monkey. EEG Clinical Neurophysiology 53, 78-93.
Kulics, A. T., C. G. Lineberry, and J. R. Roppolo (1977) Neurophysiological correlates of cutaneous discrimination performance in rhesus monkey. Brain Research 136, 360-365.
Luria, A. R. (1973) The Working Brain, Penguin, Harmondsworth, England.
Mascagni, F., A. J. McDonald, and J. R. Coleman (1993) Corticiamygdaloid and corticocortico projections of the rat temporal cortex: a Phaseolus vulgaris leucoagglutinin study. Neuroscience 57, 697-715.
Merzenich, M. M., P. L. Knight, and G. L. Roth (1975) Representation of cochlea within primary auditory cortex in the cat. Journal of Neurophysiology 38, 231-249.
Mesulam, M.-M., ed. (1982) Tracing Neural Connections with Horseradish Peroxidase, Wiley, New York.
Middlebrooks, J. C. Identification of Sounds Based on the Temporal Firing Patterns of Auditory Cortical Neurons. Symposium on the Auditory Cortex: Common Themes and Specialization for Hearing. National Society for Neuroscience Annual Meeting, 1997.
Morel, A., and T. J. Imig (1987) Thalamic projections to fields A, AI, P, and VP in the cat auditory cortex. Journal of Comparative Neurology 265, 119-144.
Morel, A., E. Rouiller, Y. de Ribaupierre, and F. de Ribaupierre (1987) Tonotopic organization of the medial geniculate body (MGB) in lightly anesthetized cats. Experimental Brain Research 69, 24-42.
Morest, D. K. (1964) The neuronal architecture of the medial geniculate body of the cat. Journal of Anatomy, London 98, 611-630.
Morest, D. K. (1965) The laminar structure of the medial geniculate body of the cat. Journal of Anatomy, London 99, 143-160.
Ojima, H. (1994) Terminal morphology and distribution of corticothalamic fibers originating from layers 5 and 6 of cat primary auditory cortex. Cerebral Cortex 4, 646-663.
Oliver, D. L., G. E. Beckius, and E. M. Ostapoff (1994) Connectivity of neurons in identified auditory circuits studied with transport of dextran and microspheres plus intracellular injection of lucifer yellow. Journal of Neuroscience Methods 53, 23-27.
Pandya, D. N. (1995) Anatomy of the auditory cortex. Revue Neurologique 151, 486-494.
Pandya, D. N., and E. H. Yeterian (1984) Architecture and Connections of Cortical Association Areas. In Cerebral Cortex Vol. 4, A. Peters, and E. G. Jones, eds., pp. 3-61, Plenum Press, New York.
Patterson, H. A. (1977) An antereograde degeneration and retrograde axonal transport study of the cortical projections of the rat medial geniculate body [Dissertation], Boston University School of Education.
Paxinos, G., and C. Watson (1982) The Rat Brain in Stereotaxic Coordinates, Academic Press, New York.
Paxinos, G., and C. Watson (1998) The Rat Brain in Stereotaxic Coodinates, 4th ed., Academic Press, New York.
Pontes, C., F. F. Reis, and A. Sousa-Pinto (1975) The auditory cortical projections onto the medial geniculate body in the cat. An experimental anatomical study with silver and autoradiographic methods. Brain Research 91, 43-63.
Raczkowski, D., I. T. Diamond, and J. Winer (1975) Organization of thalamocortical auditory system in the cat studied with horseradish peroxidase. Brain Research 101, 345-354.
Ramon y Cajal, S. (1911) Histologie du Systeme Nerveux de l'Homme et des Vertebres, L. Azoulay (trans.) ed., Maloine, Paris.
Reale, R. A., and T. J. Imig (1980) Tonotopic organization in auditory cortex of the cat. Journal of Comparative Neurology 192, 265-291.
Redies, H., S. Brandner, and O. D. Creutzfeldt (1989) Anatomy of the auditory thalamocortical system of the guinea pig. Journal of Comparative Neurology 282, 489-511.
Rioch, D. M. (1929) Studies on the diencephalon of carnivora. Journal of Comparative Neurology 49, 1-119.
Rockland, K. S., and D. N. Pandya (1979) Laminar origins and terminations of cortical connections of the occipital lobe in the rhesus monkey. Brain Research 179, 3-20.
Roe, A. W., S. L. Pallas, J. O. Hahm, and M. Sur (1990) A map of visual space induced in primary auditory cortex. Science 250, 818-820.
Roger, M., and P. Arnault (1989) Anatomical study of the connections of the primary auditory area in the rat. Journal of Comparative Neurology 287, 339-356.
Romanski, L. M., and J. E. LeDoux (1993a) Information cascade from primary auditory cortex to the amygdala: corticocortical and corticoamygdaloid projections of temporal cortex in the rat. Cerebral Cortex 3, 515-532.
Romanski, L. M., and J. E. LeDoux (1993b) Organization of rodent auditory cortex: anterograde transport of PHA-L from MGv to temporal neocortex. Cerebral Cortex 3, 499-514.
Rose, J. E. (1949) The cellular structure of the auditory region of the cat. Journal of Comparative Neurology 91, 409-439.
Rose, J. E., and C. N. Woosley (1949) The relations of thalamic connections, cellular structure and evocable electrical activity in the auditory region of the cat. Journal of Comparative Neurology 91, 441-467.
Rouiller, E. M., and F. de Ribaupierre (1985) Origin of afferents to physiologically defined regions of the medial geniculate body of the cat: ventral and dorsal divisions. Hearing Research 19, 97-114.
Rouiller, E. M., and E. Welker (1991) Morphology of corticothalamic terminals arising from the auditory cortex of the rat: a Phaseolus vulgaris-leucoagglutinin (PHA-L) tracing study. Hearing Research 56, 179-190.
Rouiller, E. M., C. Rodrigues-Dagaeff, G. Simm, Y. De Ribaupierre, A. Villa, and F. De Ribaupierre (1989) Functional organization of the medial division of the medial geniculate body of the cat: tonotopic organization, spatial distribution of response properties and cortical connections. Hearing Research 39, 127-142.
Rouiller, E. M., G. M. Simm, A. E. Villa, Y. de Ribaupierre, and F. de Ribaupierre (1991) Auditory corticocortical interconnections in the cat: evidence for parallel and hierarchical arrangement of the auditory cortical areas. Experimental Brain Research 86, 483-505.
Ryugo, D. K., and H. P. Killackey (1974) Differential telencephalic projections of the medial and ventral divisions of the medial geniculate body of the rat. Brain Research 82, 173-177.
Ryugo, D. K., and N. M. Weinberger (1978) Differential plasticity of morphologically distinct neuron populations in the medial geniculate body of the cat during classical conditioning. Behavioral & Neural Biology 22, 275-301.
Scheel, M. (1988) Topographic organization of the auditory thalamocortical system in the albino rat. Anatomy & Embryology 179, 181-190.
Uttal, W. R. (1978) The Psychobiology of Mind, Wiley & Sons, Hillsdale.
Weinberger, N. M. (1993) Learning-induced changes of auditory receptive fields. Current Opinion in Neurobiology 3, 570-577.
Weinberger, N. M. (1995) Dynamic regulation of receptive fields and maps in the adult sensory cortex. [Review] [40 refs]. Annual Review of Neuroscience 18, 129-158.
Weinberger, N. M., and D. M. Diamond (1987) Physiological plasticity in auditory cortex: rapid induction by learning. Progress in Neurobiology 29, 1-55.
Weinberger, N. M., and D. M. Diamond (1988) Dynamic Modulation of the Auditory System by Associative Learning. In Auditory Function: Neurological Bases of Hearing, G. M. Edelman, W. E. Gall, and W. M. Cowan, eds., pp. 485-512, The Neuroscience Institute (John Wiley & Sons), New York.
Winer, J. A. (1984a) Anatomy of layer IV in cat primary auditory cortex (AI). Journal of Comparative Neurology 224, 535-567.
Winer, J. A. (1984b) Identification and structure of neurons in the medial geniculate body projecting to primary auditory cortex (AI) in the cat. Neuroscience 13, 395-413.
Winer, J. A. (1984c) The non-pyramidal cells in layer III of cat primary auditory cortex (AI). Journal of Comparative Neurology 229, 512-530.
Winer, J. A. (1984d) The pyramidal neurons in layer III of cat primary auditory cortex (AI). Journal of Comparative Neurology 229, 476-496.
Winer, J. A. (1985a) The medial geniculate body of the cat. Advances in Anatomy, Embryology & Cell Biology 86, 1-97.
Winer, J. A. (1985b) Structure of layer II in cat primary auditory cortex (AI). Journal of Comparative Neurology 238, 10-37.
Winer, J. A. (1992) The Functional Architecture of the Medial Geniculate Body and the Primary Auditory Cortex. In The Mammalian Auditory Pathway: Neuroanatomy, D. B. Webster, A. N. Popper, and R. R. Fay, eds., pp. 222-409, Springer-Verlag, New York.
Winer, J. A., and D. T. Larue (1987) Patterns of reciprocity in auditory thalamocortical and corticothalamic connections: study with horseradish peroxidase and autoradiographic methods in the rat medial geniculate body. Journal of Comparative Neurology 257, 282-315.
Winer, J. A., and D. K. Morest (1983) The medial division of the medial geniculate body of the cat: implications for thalamic organization. Journal of Neuroscience 3, 2629-2651.
Winer, J. A., and J. J. Wenstrup (1994) Cytoarchitecture of the medial geniculate body in the mustached bat (Pteronotus parnellii). Journal of Comparative Neurology 346, 161-182.
Winer, J. A., I. T. Diamond, and D. Raczkowski (1977) Subdivisions of the auditory cortex of the cat: the retrograde transport of horseradish peroxidase to the medial geniculate body and posterior thalamic nuclei. Journal of Comparative Neurology 176, 387-417.
Winer, J. A., D. K. Morest, and I. T. Diamond (1988) A cytoarchitectonic atlas of the medial geniculate body of the opossum, Didelphys virginiana, with a comment on the posterior intralaminar nuclei of the thalamus. Journal of Comparative Neurology 274, 422-448.
Wong-Riley, M. (1978) Reciprocal connections between striate and prestriate cortex in squirrel monkey as demonstrated by combined peroxidase histochemistry and autoradiography. Brain Research 147, 159-164.
Woosley, C. N. (1960) Organization of cortical auditory system: a review and synthesis. In Neural Mechanisms of Auditory and Vestibular Systems, G. L. Rasmussen, and W. F. Windle, eds., pp. 201-210, Thomas, Springfield.
Woosley, C. N., and E. M. Walzl (1942) Topical projections of nerve fibers from local regions of the cochlea to the cerebral cortex of the cat. Bulletin Johns Hopkins Hospital 71, 315-344.
Yeterian, E. H., and D. N. Pandya (1989) Thalamic connections of the cortex of the superior temporal sulcus in the rhesus monkey. Journal of Comparative Neurology 282, 80-97.
Zilles, K. (1985) The cortex of the rat, Springer, New York.
Zilles, K., and A. Wree (1985) Areal and laminar structure of the rat cortex. In The rat nervous system Vol. 1, G. Paxinos, ed. Academic Press, Sydney.
Zilles, K., A. Wree, and N. Dausch (1990) Anatomy of the neocortex: neurochemical organization. In The cerebral cortex of the rat, B. Kolb, and R. C. Tees, eds., pp. 113-150, MIT Press, Cambridge.