HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (3) Citing Articles via Google Scholar Google Scholar Articles by Butler, A. B. Articles by Cotterill, R. M. J. Search for Related Content PubMed PubMed Citation Articles by Butler, A. B. Articles by Cotterill, R. M. J.

Mammalian and Avian Neuroanatomy and the Question of Consciousness in Birds

¹ The Krasnow Institute for Advanced Study and Department of Psychology, George Mason University, Fairfax, Virginia
² Biophysics Group, Danish Technical University, DK-2800 Lyngby, Denmark

^* To whom correspondence should be addressed, at Krasnow Institute for Advanced Study, MSN 2A1, George Mason University, Fairfax, VA 22030. E-mail: abbutler{at}gmu.edu

	Abstract

TOP Abstract Introduction The Complex Cognitive Abilities... Cognition, Consciousness, and... Similarities and Differences in... Reptilian and Amphibian... Conclusions Appendix Glossary of Terms Pertaining... Literature Cited

 
Some birds display behavior reminiscent of the sophisticated cognition and higher levels of consciousness usually associated with mammals, including the ability to fashion tools and to learn vocal sequences. It is thus important to ask what neuroanatomical attributes these taxonomic classes have in common and whether there are nevertheless significant differences. While the underlying brain structures of birds and mammals are remarkably similar in many respects, including high brain-body ratios and many aspects of brain circuitry, the architectural arrangements of neurons, particularly in the pallium, show marked dissimilarity. The neural substrate for complex cognitive functions that are associated with higher-level consciousness in mammals and birds alike may thus be based on patterns of circuitry rather than on local architectural constraints. In contrast, the corresponding circuits in reptiles are substantially less elaborated, with some components actually lacking, and in amphibian brains, the major thalamopallial circuits involving sensory relay nuclei are conspicuously absent. On the basis of these criteria, the potential for higher-level consciousness in these taxa appears to be lower than in birds and mammals.

	Introduction

TOP Abstract Introduction The Complex Cognitive Abilities... Cognition, Consciousness, and... Similarities and Differences in... Reptilian and Amphibian... Conclusions Appendix Glossary of Terms Pertaining... Literature Cited

 
Elucidating the mechanism or mechanisms by which sets of neurons generate consciousness requires that one be able to identify at least some of the animals, in addition to humans, that experience it. Different levels of consciousness have been variously defined, the most basic of which is referred to as "primary consciousness" (Edelman, 1989, 1992) or "perceptual consciousness" (Pepperberg and Lynn, 2000) and is limited to the experience of having mental images in the present. The more complex levels of consciousness build on primary consciousness, or basic sensory awareness, to reach, for example, Edelman’s (1989, 1992) "higher-order consciousness," which involves a sense of one’s own self and an awareness of being aware, or Pepperberg and Lynn’s (2000) "reflective consciousness," which includes awareness of being aware and additionally comprises high-level cognitive abilities such as possession of a theory of mind, the ability to practice deception, and executive-type decision making. In this paper, we will discuss higher levels of consciousness that encompass many complex (higher-order) cognitive activities but may or may not include self-awareness and being aware that one is aware, and we therefore will use the term "higher-level consciousness" (HLC), rather than "higher-order consciousness."

A basic premise of the neuroscientific study of consciousness—at any or all levels—is that it is a biological phenomenon that occurs as the result of specific neural activity, as discussed by Butler et al. (2005). It is probable that primary, or perceptual, consciousness is widely distributed among vertebrates and possibly some or many invertebrate organisms. However, in the absence of an understanding of the neural mechanism of its generation (beyond raw perception and reflex), it is currently impossible to determine whether that is so. We posit here that higher levels of consciousness can be recognized, in at least some of the taxa in which they occur, on the premise that since HLC is correlated with complex cognitive abilities in humans, it is biologically parsimonious to assume that such a correlation holds for other animals as well.

While the subjective experience of consciousness, at whatever level, cannot be studied directly, its occurrence in a variety of nonhuman animals is being addressed (Baars, 2005). In nonhuman mammals, a broad range of criteria can be employed in comparative studies, including EEG patterns, cortical and thalamic integrity and activity, widespread brain activation, and various attributes of the stimulus or its content, as discussed by Seth et al. (2005), who argue for the occurrence of at least primary consciousness in nonhuman mammals, dependent on the thalamocortical complex. Further, Edelman et al. (2005) consider the possibility of some level of consciousness in some nonmammalian species, such as birds. They note that while birds lack the neocortical structure of mammals, they do have homologous basal ganglia and also may have "structural and functional equivalents" of the mammalian thalamocortical system. They acknowledge that in addition to the likely occurrence of primary, or perceptual, consciousness in birds, some of the highly complex behaviors of some taxa, particularly parrots, such as being able to make discriminations about discriminations (e.g., identifying changes in an array of objects [Pepperberg and Shive, 2001]), may indicate a level of consciousness in the higher-level range. Likewise, Emery and Clayton (2005) have discussed the evolution of intelligence and higher-level cognitive functions in birds in the context of the similarities in neural connectional patterns and associational pallial areas with mammals. The present paper is an attempt to explore in much greater detail the possible shared neural features of birds and mammals that may generate such higher levels of consciousness.

In terms of behavioral correlates of higher-order (or higher-level) consciousness, Baars (2003) argues that working memory, which has been described for humans and other primates by Fuster (2003, p. 155) as "active memory for the short term," requires consciousness, rather than the other way around. In primates, including humans, working memory is dependent upon the integrity of the prefrontal cortex and is necessary for many higher-order cognitive functions (Goldman-Rakic, 1996; Fuster, 2003), specifically those that require use of "a cognit, new or old, which is held active in the focus of attention as required in the processing of information for prospective action" (Fuster, 2003, p. 155). Working memory is essential for delayed-response tasks, and in humans, it makes a crucial contribution to syntactic construction, language comprehension, and reading and writing (Fuster, 2003). In discussing consciousness, Fuster (2003, p. 251) states that "Working memory ... is the internalized attention on a recent percept for prospective action, and thus the persistent activation of the cognitive network that represents that percept. Memory can enter consciousness in a multitude of forms and states. The recall of any memory is conscious by definition," as is also imagining. Fuster argues that perception, memory, and, most of all, attention contribute to consciousness, and (2003, p. 253) "When a task imposes demands on the brain’s capacity to integrate information across time, attention becomes arrested in the activation and retention of time-integrating cognits, as these become the content of working memory."

An important caveat must be noted here, which is that higher-level consciousness cannot be ruled out in organisms that do not display complex cognitive abilities. Nonetheless, where such abilities are present, it seems reasonable to suggest that the subjective phenomenon of higher-level consciousness is likewise present. Accordingly, we agree with Århem and Liljenstrom (1997) that the evolution of consciousness and its degree appear to be associated with the evolution of increased complexity in the central nervous systems of some taxa (see Butler and Hodos, 2005).

The posited correlation of complex cognition with higher-level consciousness, even with any level of consciousness at all, has been disputed. For example, Libet (2004) notes that highly complex cognitive activities, such as solving a mathematical problem or creative thinking, can be done at the subconscious level in humans and, thus, complex problem solving in other animals may also proceed at an unconscious level rather than a conscious one. While the latter is true and probably would require a specific set of neural design constraints in terms of circuitry, we dispute this argument for the exclusivity of higher-level consciousness to humans for three main reasons. First, in biological terms, it is not parsimonious to suggest that the higher-level consciousness experienced by normal humans occurs in one and only one animal species, particularly since no entirely unique neural feature (i.e., any qualitative difference from other primates or mammals in general) has yet been identified in humans (Preuss, 2000; Hof et al., 2001; Roth and Dicke, 2005). Second, the correlation of degrees of consciousness with degrees of neural complexity and their coevolution (Århem and Liljenstrom, 1997) is a likewise biologically parsimonious postulate. Third, Baars’s (2003) argument for the evolution of working memory in primates and the dependence of its higher-level conscious components upon primary consciousness (i.e., spontaneous sensory imagery) is testable across nonprimate animals, including birds, at the behavioral level through measurements of delay and other tasks associated with working memory.

A productive strategy to elucidate the neural mechanisms of the components of higher-level consciousness, which include primary consciousness, is to use the comparative approach to identify what neural features are shared among divergent organisms exhibiting complex cognitive abilities, and thus presumably experiencing higher-level consciousness. Once the mechanisms for the neural generation of the varying levels of consciousness have been identified, one could then examine which of these mechanisms are operative across a wide variety of vertebrate and invertebrate taxa.

In brief, our strategy is based on the following reasoning: (1) Because the occurrence of HLC and complex cognition are correlated in humans, they are presumed to be similarly correlated in other animals. (2) In humans and other mammals, the main neural structures that have been postulated to produce HLC (e.g., Cotterill, 1997; Tononi and Edelman, 1998; Crick and Koch, 2003, 2005; Edelman, 2003;) include the claustrum, most or all of the neocortex, the dorsal thalamic specific sensory relay nuclei, the intralaminar nuclei, and the thalamic reticular nucleus. (3) Birds exhibit highly complex cognitive abilities and thus also may share the experience of HLC with humans and other mammals. (4) If birds and mammals both experience HLC, only those neural features that are common to both would be those that are crucial for its production. (5) If birds and mammals both experience HLC, proposed models for the neural production of consciousness (e.g., Edelman’s dynamic core model [Tononi and Edelman, 1998]) should be consistent with the identified shared neural features, both in terms of anatomical traits and electrophysiological correlates of observed behaviors. Thus, we first will review some of the complex cognitive abilities exhibited by various species of birds and then survey which potentially relevant neural features are common to both birds and mammals.

	The Complex Cognitive Abilities of Birds

TOP Abstract Introduction The Complex Cognitive Abilities... Cognition, Consciousness, and... Similarities and Differences in... Reptilian and Amphibian... Conclusions Appendix Glossary of Terms Pertaining... Literature Cited

 
An increasing and now substantial body of evidence indicates that birds not only have some cognitive abilities, but that at least some species have impressive ones (Delius et al., 2001; Güntürkün and Durstewitz, 2001; Macphail, 2001). For example, transitive inference is regarded as a cognitively challenging task in which the subject must infer relationships across stimuli when presented serially with multiple bits of information. This phenomenon has been demonstrated in studies of social structure formation in great tits (Parus major), which are parid songbirds, by Peake et al. (2002). In this study, the behavior of male great tits was assessed in a territorial intrusion sequence simulated by using loudspeakers placed inside and outside the territorial boundary of the subject (S). The degree of aggression of the first simulated intruder (A) was varied by the degree of temporal overlap of song from the loudspeaker with that produced by S: the more aggressive condition was concurrent playing of the stimulus song with S’s song, and the less aggressive condition was starting the stimulus song 1 second following S’s song. A simulated interaction between A and another unknown individual, B, then occurred outside the territory of S, on which S could eavesdrop, followed by intrusion by B into S’s territory. The subjects were able to combine information about the first encounter with A and A’s subsequent encounter with B to determine their own level of response to B. For example, in terms of aggression, if A>S and A<B, then B>S. The responses of the subjects in terms of their own song output to the intrusion of B were consistent with being able to solve this paradigm. Similarly, pinyon jays (Gymnorhinus cyanocephalus), which are highly social corvids, determine their own relative rank within the hierarchy by combining information from within-group direct dominance encounters (competing with another individual for access to a peanut) with information gained from subsequently observing encounters between other birds (Paz-y-Miño et al., 2004).

Another phenomenon called coherence, or multistability, of ambiguous figures has been considered unique to primates, but pigeons likewise exhibit it. This phenomenon is thought to require mental expectation, and it involves a switching of perception for figures such as the Necker cube, for which the perceived open end alternates in sets of trials between the two possibilities. Pigeons exhibit the same type of alternation for perception of either horizontal or vertical movement (Vetter et al., 2000). Episodic memory, once thought to be unique to humans, is also within the capabilities of birds. Scrub jays (Aphelocoma coerulescens) demonstrate it for the location, content, and relative time since caching particular food items (Clayton and Dickenson, 1998, 1999; Clayton et al., 2003). Episodic memory has also been convincingly demonstrated in pigeons by Zentall et al. (2001). The experiment was designed not only to ask whether the subjects had memory of a past experience but to ask it in an unexpected way, such that associational learning or anticipation could not be employed by the subjects in their responses to the text paradigm.

Object constancy requires a high level of cognitive complexity. It has been demonstrated to Piagetian stage 4a (constancy of a disappeared object if a grasping or approach movement toward it has been initiated at the time of disappearance) in ring doves (Dumas and Wilkie, 1995) and to stage 5 (constancy over sequential, multiple disappearances) in magpies (Pollok et al., 2000). African grey parrots (Psittacus erithacus) and other psittacine birds demonstrate object constancy to Piagetian stage 6, which includes the ability to follow complex displacements of a disappeared object, as well as to form a mental representation of the object. Parrots behave in a manner consistent with surprise and anger when an unexpected, different object is used to surrepetitiously replace the disappeared object (Pepperberg et al., 1997).

Tool manufacture, long regarded as an exclusive province of primates, has now also been observed in a corvid species. New Caledonian crows (Corvus moneduloides) fabricate tools for use in gaining access to food, both in the wild and in captivity. In the wild, these crows have developed specific designs in their tool manufacture using Pandanus leaves, which are then shared with conspecifics through cultural transmission (Hunt and Gray, 2003). In captivity, members of this species have been filmed (Weir et al., 2002) while bending pieces of wire into hooks, which they used to lift food-laden baskets from the bottom of a long glass tube, the tube being much longer than the birds’ beaks and necks (Fig. 1). Recently, these observations were augmented by the finding that hand-raised, naive juvenile crows of this species spontaneously used available twigs as tools to retrieve otherwise unreachable food items and, further, when presented with Pandanus leaves, spontaneously used them to manufacture tools for this same use (Kenward et al., 2005).

View larger version (104K): [in this window] [in a new window]   Figure 1. A New Caledonian crow (Corvus moneduloides) using a tool that she manufactured by bending a piece of wire to retrieve a bucket that contains food. Picture reproduced with the kind permission of Alex Weir, Jackie Chappell, and Alex Kacelnik, Dept. of Zoology, University of Oxford.

Finally, theory of mind, which involves attribution of one’s own mental state and possible future behavior to another organism and has been considered to be exclusive to primates, also may be within the cognitive capabilities of some birds. Caching behaviors and strategies indicative of theory of mind have been reported by Emery and Clayton (2001) and Clayton et al. (2003): scrub jays that had previously cached food items while observed by another jay were given an opportunity to remove and recache the food without being observed; only jays who themselves had stolen from other birds recached their own food items in this situation. Likewise, ravens have been found to engage in deliberately deceptive maneuvers designed to lure a dominant conspecific away from the concealed food (Bugnyar and Kotrschal, 2002, 2004).

The range and level of complexity of the above behaviors of various taxa of birds indicate high levels of cognitive ability. As noted above, the postulate that such ability is correlated with and indeed supported by higher-level consciousness thus leads to the hypothesis that birds and mammals share that level of consciousness. Such findings, a number of which recently have been reviewed in depth by Emery and Clayton (2004), give renewed impetus to the debate concerning possible subjective consciousness in birds (Walker, 1983; Radner and Radner, 1989; Dawkins, 1993; Trefil, 1997; Griffin, 2001). Birds meet the cognitive criteria for having working memory (Diekamp et al., 2002) that, as discussed above, Baars (2003) posits as requiring consciousness. These observations suggest that one should ask what neural features birds and mammals share. If species belonging to these two taxonomic classes share at least some aspects of higher cognition, and thus higher-level consciousness, they might be expected also to share the neural features that generate those capacities.

	Cognition, Consciousness, and the Neural Bases of Complex Functions

TOP Abstract Introduction The Complex Cognitive Abilities... Cognition, Consciousness, and... Similarities and Differences in... Reptilian and Amphibian... Conclusions Appendix Glossary of Terms Pertaining... Literature Cited

 
In their review of the evolution of intelligence, Emery and Clayton (2004) note that "divergent brain evolution" has resulted in substantial differences across amniotes in brain organization at the cellular level. They cite the "vastly different brain structures" but do not discuss them, nor do they explore the similarities that nonetheless exist. The present paper compares the neural features that have been proposed to generate complex cognition and consciousness in mammals (Cotterill, 2001) with the neural features present in avian brains. As we will discuss, it is in the circuitry of mammalian and avian brains, rather than in their cytoarchitecture, that a marked degree of similarity exists. It appears equally pertinent to ask whether there are neural features that mammalian brains and avian brains do not share. It is hoped that such a comparison will further the understanding of consciousness at the neuronal level.

A recent analysis (Cotterill, 2001) tentatively identified the brain components critical for consciousness in mammals and suggested how it is generated. This analysis maintains that the traditional stimulus-response paradigm of consciousness is fundamentally flawed, and it suggests that significant progress will not be achieved unless that paradigm is replaced by one that accords primacy to sequences of muscular movements—actual or actively prepared for—and the resulting sensory feedback from the surroundings— actual or actively anticipated. It proposes that the evolutionary advantage of consciousness lies in its permitting an individual member of a species to acquire novel context-specific reflexes within its own lifetime. Indeed, it holds that consciousness is a prerequisite for such acquisition.

The analysis postulates that mammals learn about their surroundings primarily by probing them with muscular movements, and by detecting and experiencing the resulting sensory feedback. It focuses on sequences of muscular movements rather than on individual muscular flexions because the latter have only limited cognitive potential; one can learn very little about an unseen object merely by tapping it, but much more when allowed to explore it with a sequence of hand and finger movements. The muscular movements are generated internally, by a drive mechanism (probably centered on the basal ganglia in mammals) that seeks to maintain the internal status quo—that is to say, by homeostasis. The conjectured consciousness mechanism involves dispatch to the sensory receptors of alerting signals—efference copy signals (Cotterill, 1997)—emanating from the brain region responsible for generating sequences of muscular movements, particularly the premotor cortex and the supplementary motor cortex.

The central idea of this new theory (Cotterill, 2001) is therefore that sequences of muscular movements ask questions of the surroundings, and that the sensory receptors receive the answers. It postulates an "exploration-information" paradigm rather than a "stimulus-response" one. Correct pairing of answer to question requires a certain span of working memory, which in mammals is generally taken to implicate parts of the frontal lobe. Questions must be correctly directed if they are to be of value, so the system needs to be able to focus its attention when exploring its surroundings. The theory is thus novel in suggesting that the primary role of attention lies in the selection of questions put to the surroundings, rather than in selecting from among the ensuing signals of sensory feedback. The analysis (Cotterill, 2001) concludes that the attention mechanism involves the thalamic intralaminar nuclei and the nucleus reticularis thalami.

The theory thus sees key roles in consciousness being played by the premotor and supplementary motor cortices, the prefrontal cortex, the basal ganglia, the sensory cortices, and the various dorsal thalamic nuclei. In its focus on that set of neural components, particularly the premotor and supplementary motor cortices, this theory differs from other current theories of consciousness in mammals (e.g., John, 2001; Roth, 2001; Crick and Koch, 2003; Edelman, 2003; Freeman, 2003; Osaka, 2003). An important question that is rarely addressed, however, is whether these neural components in mammals are requisite in their entirety for the production of consciousness or whether a subset of their features can produce it. A comparative approach is required for addressing this question. In what follows, therefore, we will examine to what extent the features of these neural components are found both in the brains of mammals and in the brains of birds.

	Similarities and Differences in Mammalian and Avian Brain Organization: Keys to Cognition and Consciousness?

TOP Abstract Introduction The Complex Cognitive Abilities... Cognition, Consciousness, and... Similarities and Differences in... Reptilian and Amphibian... Conclusions Appendix Glossary of Terms Pertaining... Literature Cited

 
The comparisons we are able to make here are based on the new understanding of the avian brain that has come about during the past several decades, due to the pioneering studies of Harvey Karten that began in the 1960s and culminated with a new nomenclature for many of the structures in the bird forebrain (Reiner et al., 2004a, b; The Avian Brain Nomenclature Consortium, 2005). See Appendix for an overview, including a glossary of terms, of forebrain organization in amniotes. As shown in Figure 2, the telencephalons of both birds and mammals have a large region of pallium (shaded in medium gray), which is the upper part of the telencephalon developmentally, and which overlies the subpallium (shaded in darker gray).

View larger version (66K): [in this window] [in a new window]   Figure 2. Nissl hemisections with mirror-image drawings through the telencephalons of (A) a mouse and (B) a pigeon. The section through the mouse brain is somewhat more caudal than that of the bird to show the location of the amygdala. The pallium is indicated with lighter shading and the subpallium with darker shading.

In both mammals and birds, the most medial part of the pallium is the hippocampal formation, which is the major pallial component of the limbic system and is involved in spatial memory and other learning functions (Bingman and Able, 2002; Vargas et al., 2004). Likewise, an olfactory cortex is present in both birds and mammals. In mammals, the remaining and largest part of the pallium is the neocortex, which is composed of six horizontally aligned cell and fiber layers and comprises millions of vertically organized, functional columns of cells and their synaptic interconnections.

The neocortex can be divided into two parts on the basis of the pattern of its thalamic afferentation—a lateral, or collocortical, part that receives its predominant input from the collothalamus, the set of nuclei that receive their predominant input via a relay through the midbrain roof; and a medial, or lemnocortical, part that receives its predominant input from the lemnothalamus, the set of nuclei that receive their predominant inputs more directly and without relay through the roof of the midbrain (Butler, 1994a, b; 1995). Currently available evidence clearly indicates that while significant developmental differences exist (Medina and Reiner, 2000), the hyperpallial part of the avian telencephalon, also referred to as the Wulst, is homologous, as a region, to the medial, lemnocortical part of mammalian neocortex (Butler, 1994b; Reiner, 2000).

In contrast, the evolutionary relationship of the rest of the avian pallium—including the mesopallium and the nidopallium—is the subject of controversy. The nidopallium in particular contains several subregions that receive ascending, collothalamic sensory inputs. The entopallium, which is shown in Figure 2, is one such subregion, and it receives the ascending visual inputs relayed from the optic tectum (the avian counterpart of the mammalian superior colliculus) through the dorsal thalamus. Whether these various parts of the avian pallium are homologous to the neocortex, to the claustrum-endopiriform formation, or to the basolateral part of the pallial amygdala (all of which receive projections from collothalamic nuclei), to a combination of one or more of these structures, or to none of them remains unresolved (Bruce and Neary, 1995; Butler and Molnár, 2002; Northcutt and Kaas, 1995; Karten, 1997; Martínez-Garcia et al., 2002; Puelles et al., 2000; Reiner, 2000).

For purposes of the comparisons being made here, the homology of these lateral pallial structures is less important than is their comparable place within the circuitry of the forebrain. Multiple major ascending sensory pathways to the pallium, multiple descending motor pathways to the spinal cord, and a number of circuit loops that share a common pattern of pallio-striato-pallido-thalamo-pallial projections are present in both birds and mammals. This circuitry thus appears to be a fundamental feature that contributes to the production of complex cognitive functions and possibly to the associated experience of higher-level consciousness.

Mammalian brain organization
Figure 3 shows the anatomical connections between the various regions of the mammalian brain. It is a modified version of a figure published earlier (Cotterill, 2001), the alterations facilitating comparison with the connections in the avian brain (see Fig. 4). The sensory areas of the neocortex are divided into lemnocortex and collocortex at the upper right, which together include the occipital, parietal, and temporal cortical areas; the lemnothalamic-recipient prefrontal and motor-related neocortical areas are shown separately at the upper left. No attempt has been made to indicate the rich multiplicity of areas found in the neocortex’s occipital, parietal, and temporal lobes (Felleman and Van Essen, 1991).

View larger version (46K): [in this window] [in a new window]   Figure 3. Diagram of the major sensory and motor circuits in mammals, modified from Cotterill (2001). Neocortical regions are indicated in gray, as are the claustrum and the basolateral part of the pallial amygdala; the thalamic nuclei are shown in blue; and the basal ganglia and cerebellum in yellow. The collicular components of the midbrain roof are colored aquamarine, the brainstem motor relay nuclei orange, and the sensory receptors and first-order relay cell populations magenta. Excitatory inputs are indicated by green arrows, and most inhibitory inputs by red arrows. Excitatory inputs from numerous neocortical areas to the thalamic reticular nucleus are indicated by a large green arrowhead at its upper right border, and inhibitory input to dorsal thalamic nuclei from the nucleus reticularis thalami is indicated for each by a red arrowhead at its upper left border. Figure modified and reprinted from Progress in Neurobiology, Vol. 64, R. M. J. Cotterill, Cooperation of the basal ganglia, cerebellum, sensory cerebrum and hippocampus: possible implications for cognition, consciousness, intelligence and creativity, pp. 1–33, Copyright 2001, with permission from Elsevier.

View larger version (44K): [in this window] [in a new window]   Figure 4. Diagram of the major sensory and motor circuits in birds (references in text and Arends and Zeigler, 1991; Boord, 1968; Casini et al., 1986; Clarke, 1977; Cowan et al., 1961; Crowder et al., 2000; Deng and Rogers, 2000; Husband and Shimizu, 1999; Jiao et al., 2000; Karten, 1967, 1968; Karten and Hodos, 1970; Karten and Revzin, 1966; Kroner and Güntürkün, 1999; Medina and Reiner, 1997; Miceli et al., 1987, 1990; Necker, 2001; Ovalle et al., 1999; Shimizu et al., 1995; Vates, 1997; Veenman et al., 1995; Wild, 1987, 1989, 1992, 1997; Wild and Williams, 1999; Winship and Wylie, 2003; Yamamoto et al., 2000; see also Butler and Hodos, 2005) for comparison with that of mammals. For ease of comparison with Figure 3, the non-olfactory and non-hippocampal pallial regions have been colored gray, the thalamic nuclei blue, and the basal ganglia and cerebellum yellow. The collicular components of the midbrain roof are aquamarine (and labeled with mammalian terms), the brainstem motor relay nuclei orange, and the sensory receptors and first-order relay cell populations magenta. Excitatory and inhibitory inputs generally are indicated as in Figure 3 but with open red arrowheads used for possible inhibitory input to some dorsal thalamic nuclei from nucleus reticularis thalami (usually called the superior reticular nucleus in birds), since such input has not yet been established experimentally except for nuclei rotundus and ovoidalis (Mpodozis et al., 1996). The excitatory input from various pallial areas to the thalamic reticular nucleus is indicated by a large green arrowhead at its upper right border.

The latter, nonintralaminar systems have not been included in Figures 3 or 4, since they are widely distributed among vertebrates (see Butler and Hodos, 2005) and believed to be involved in regulating the level of neural activity rather than in the production or experience of its subjective content per se.

The specific contributions of those structures involved in maintaining wakefulness to the subjective experience of consciousness cannot be determined for the present. However, a number of additional structures within the forebrain have been posited to be involved in or essential for consciousness—at the primary or possibly some higher level—in mammals. One of these structures is the claustrum. Lying deep to the main cortical layers in the lateral part of the cerebral hemisphere and receiving ascending input from the intralaminar nuclei, the claustrum might or might not be a good candidate for participation in the generation of consciousness in mammals. Crick and Koch (2005) recently examined the possible role of the claustrum in consciousness, suggesting that, due to the global rather than point-to-point nature of its reciprocal connections with neocortex, it may play a role in orchestrating coordination in neural activity among sets of cortical neurons in disparate locations. Among monotremes, a claustrum is absent altogether in the platypus, and whether it is present in echidnas is currently unresolved (Butler et al., 2002; Ashwell et al., 2004). Thus, since monotremes, including both the platypus and the echidna, are capable of maintaining a waking state comparable to that of other mammals, as confirmed by electrophysiological evidence (Siegel et al., 1996, 1999), structures other than the claustrum, such as the reticular activating system, must be sufficient to support this ability. In placental mammals, the claustrum may thus contribute to consciousness without being essential for it.

In contrast to the claustrum with its more global projection patterns, the thalamic sensory relay nuclei for the visual, auditory, somatosensory, motor feedback, and other systems project to specific and restricted regions of the cortex. These thalamic nuclei, along with the intralaminar nuclei and the thalamic reticular nucleus, which is discussed below, are thought to be essential for higher levels of consciousness (e.g., Steriade et al., 1996; Llinás et al., 1998, 2005; Steriade, 2006) In Figure 3, the main collothalamic pathways are shown as relays from the colliculi to the thalamus; the visual and somatosensory/multisensory pathways are shown as running together from the superior colliculus, and the main auditory pathway is shown as running from the inferior colliculus. Two of the main lemnothalamic pathways are shown from the dorsal column nuclei to the ventral posterolateral nucleus (VPL) for the somatosensory system and from retinal ganglion cells directly to the dorsal lateral geniculate nucleus (DLGN) for the visual system. These various thalamic nuclei then project to their respective areas of the pallium and receive reciprocal projections in return.

The thalamocortical and reciprocal corticothalamic circuits are regulated by the nucleus reticularis thalami, also known as the thalamic reticular nucleus (TRN). As each thalamic nucleus projects to part of the cortex, it supplies part of the TRN with a glutamatergic, excitatory collateral. Each cortical area projects reciprocally back to its thalamic afferent source, and it likewise provides glutamatergic collaterals to that same part of the TRN. The TRN, which contains GABAergic, inhibitory neurons, then innervates the same thalamic afferent site as well as additional thalamic sites. Thus, the positive feedback loop between the thalamus and the cortex is kept in check and at an appropriate level of activity by the inhibitory TRN input. The TRN can additionally inhibit multiple thalamic regions, thus possibly playing a role in focusing attention by preventing the transmission of distracting stimuli (Yingling and Skinner, 1977; Guillery et al., 1998). This circuitry, in turn, emphasizes the importance of the projections to the TRN, which are particularly dense when they emanate from the anterior cingulate cortex (Cornwall et al., 1990; Lozsádi, 1994).

GABAergic, inhibitory neurons are now viewed as playing a crucial role in generating oscillatory acitivity in the gamma band (30–50 Hz) within the thalamocortical system, which is characteristic of the EEG during activated states–both waking and paradoxical (REM) sleep. The GABAergic neurons within the neocortex, the TRN, and also within thalamic sensory relay nuclei (Golshani et al., 2001) affect the firing patterns of pyramidal projection cells to produce synchronization in this range (Steriade et al., 1996; Llinás et al., 1998, 2005; Steriade, 2001, 2006). Llinás et al. (2005) have recently proposed a model of thalamocortical circuit dynamics that addresses how these sets of GABAergic neurons produce coordination in projection neuron firing patterns, viewed as crucial for the generation of temporal binding and the experience of consciousness (Llinás et al., 1998).

Of the basal ganglia, the dorsal striatum (or caudate nucleus and putamen) is indicated in the figure solely by the label "striatum"; part of the ventral striatum is represented by the nucleus accumbens; and the dorsal pallidum (that is to say, the globus pallidus) is shown. The ventral pallidum is omitted from the figure, since its projections parallel those of the dorsal pallidum. The dorsal and ventral striatopallidal complexes each project to part of the dorsal thalamus and are involved in similar circuitry, with the dorsal striatopallidal complex being more involved in somatic functions and the ventral complex more in limbic-related ones. Here, the projections of the globus pallidus to the ventral anterior and ventral lateral (pars oralis) thalamic nuclei, VA and VLo, are shown, as are the globus pallidal projections to the intralaminar nuclei. The latter nuclei were discussed above, and in addition to their widespread cortical projections, they also project directly back to the striatum, completing a thalamo-striato-pallido-thalamic loop.

VA and VLo project to motor-related areas of the frontal lobe. The dopaminergic inputs of the substantia nigra to the striatum are also shown, as are the interconnections of the globus pallidus with the subthalamic nucleus. The so-called direct loop of cortex to dorsal striatum to globus pallidus to dorsal thalamus and back to cortex is a circuit that promotes the initiation of movement, while a second circuit that additionally involves the subthalamic nucleus, the so-called indirect loop, inhibits movement (Büchel et al., 1999; Graybiel, 2000; Kayahara and Nakano, 1996; Nakano et al., 1996; Hoover and Strick, 1999). Projections from the cerebellum are relayed to separate divisions of the ventral lateral nucleus, a major role of the cerebellum being to guide movements, once they have been initiated by the action of the direct loop and the suppression of the indirect loop by dopaminergic inputs to the dorsal striatum (Hoover and Strick, 1999; Middleton and Strick, 1998). The structure of the dorsal striatum is heterogeneous (Graybiel and Ragsdale, 1978), consisting of a continuous matrix dotted by isolated patches known as striosomes. These two components are indicated in Figure 3 by the letters M and S. This dual structure has been conjectured to underlie the multiplicity of action/thought options that are clearly available to humans and are possibly employed by all mammals (Cotterill, 1998, 2001). The striosomes are the major source of GABAergic projections to the dopaminergic neurons of the substantia nigra pars compacta.

Each of the circuit loops in mammals thus can be assigned to one of three categories. First, the intralaminar-basal ganglia loop category is the shortest, since it does not involve the cortex; this is the thalamo-striato-pallido-thalamic, or TSpT, type of loop. Second, there is a set of circuits that involve only one or a few cortical (pallial) areas and can be summarized as thalamo-pallio-striato-pallido-thalamic, or TPSpT, loops. Finally, some of the cortical circuits involve connections of many different cortical areas before providing input to the striatopallidal complexes and can be summarized as thalamo-pallio-pallio-striato-pallido-thalamic, or TPPSpT, loops. These circuits all share two basic features: their glutamatergic, excitatory dorsal thalamic and pallial projection neuronal components exhibit spontaneous activity, so that even without incoming stimulation, some activity is maintained, and the GABAergic neurons of the striatopallidum and thalamic nuclei exert an inhibitory, regulatory influence that intervenes in the otherwise positive feedback loops. Some or all of these categories of circuits, varying in their number of synapses and thus in their complexity, may contribute to complex cognition and consciousness in mammals.

Avian brain organization
Figure 4 shows the anatomical connections between the various regions of the avian brain. Many of the major pathways and circuits present in mammalian brains and identified by various workers as crucially involved in the generation and maintenance of consciousness are also present in avian brains. The sensory areas of the pallium are divided into the lemnopallial Wulst and collopallial anterior dorsal ventricular ridge (mesopallium and nidopallium) at the upper right, while the lemnothalamic-recipient motor (and possibly premotor) Wulst is shown separately at the upper left. Also at the upper left are two regions that have been proposed as candidates for a homologue of mammalian prefrontal cortex—a collopallial region called the nidopallium caudolaterale (Waldmann and Güntürkün, 1993; Hartmann and Güntürkün, 1998; Diekamp et al., 2002; Lissek et al., 2002; Rose and Colombo, 2005) and a lemnopallial region called the dorsolateral corticoid area (Montagnese et al., 2003). The latter region has more recently been proposed as a homologue of mammalian cingulate cortex (Atoji and Wild, 2005), which is a neocortical component of the limbic system and related to prefrontal cortex in terms of its lemnopallial nature and its connections. Since the correspondence of any single area in the avian brain to mammalian prefrontal cortex has thus not been established, these two regions—the nidopallium caudolaterale and the dorsolateral corticoid area—are shown in a slightly different position in Figure 4 than that of the prefrontal cortex in Figure 3. Nonetheless, the involvement of these two regions in complex cognitive functions is strongly indicated. As in mammals, there are multiple interconnections of these and other pallial regions that have not been included here.

A number of different thalamic nuclei project to these various parts of the pallium (Reiner et al., 2005). A set of nuclei that are homologous to the mammalian intralaminar nuclei are present in the same part of the thalamus and have similar pallial and striatopallidal circuitry. They are identified as part of the dorsal thalamic zone, or DTZ (Veenman et al., 1997). Like the mammalian intralaminar nuclei, the intralaminar components of the DTZ also receive pallidal projections and project directly to the striatum, completing a thalamo-striato-pallido-thalamic loop. This is the TSpT type of loop invoked above.

In contrast, and as in mammals, the thalamic sensory relay nuclei for the visual, auditory, somatosensory, motor feedback, and other systems project to specific and restricted regions of the pallium. In Figure 4, the main collothalamic pathways are shown as relays from the colliculi (midbrain roof) to the thalamus; the visual and somatosensory/multisensory pathways are shown as running together from the superior colliculus, and the main auditory pathway is shown as running from the inferior colliculus. The collothalamic nuclei project to parts of the nidopallium. Two of the main lemnothalamic pathways are shown from the dorsal column nuclei to the nucleus dorsalis intermedius ventralis anterior (DIVA) and from retinal ganglion cells directly to the dorsal lateral geniculate nucleus (DLGN), a nucleus previously referred to in birds as the nucleus opticus principalis thalami, or OPT. These various lemnothalamic nuclei then project to their respective areas of the Wulst. Reciprocal projections from the sensory pallium are less well developed in birds, but have been demonstrated for both the visual and somatomotor parts of the Wulst to their respective dorsal thalamic relay nuclei (Karten et al., 1973; Wild and Williams, 2000).

It is likely that birds have some thalamopallial and reciprocal palliothalamic circuits that involve the homologue of the mammalian nucleus reticularis thalami. In birds, this nucleus (previously called the pars dorsalis of the superior reticular nucleus), as in mammals, contains GABAergic neurons (Veenman and Reiner, 1994; Sun et al., 2005), receives an input from the globus pallidus (Medina et al., 1997), and projects to various parts of the dorsal thalamus (Benowitz and Karten, 1976; Mpodozis et al., 1996). That birds are able to focus their attention on salient stimuli is clearly apparent from the cognitive literature, but whether they have all the components of the pallial-TRN-thalamic circuitry present in mammals is not yet known. Birds have GABAergic interneurons throughout the pallium, including within the thalamorecipient sensory areas (Sun et al., 2005), but in contrast to mammals, the major sensory relay nuclei, including nuclei rotundus and ovoidalis and the dorsal lateral geniculate nucleus, are only very sparsely populated with GABAergic neurons (Mpodozis et al., 1996; Sun et al., 2005).

In Figure 4, the avian dorsal striatum is indicated merely by the label "striatum," while part of the ventral striatum is represented by the nucleus accumbens, as also in Figure 3 for mammals. The dorsal pallidum, the globus pallidus, is shown, and as for mammals, the ventral pallidum is omitted from the figure. In birds, the globus pallidus has only one segment, in contrast to the two segments present in mammals, but it contains the same complement of different cell populations. As in mammals, the dorsal and ventral pallida each project to part of the dorsal thalamus and are involved in similar circuitry, with the dorsal striatum and pallidum being more involved in somatic functions and the ventral striatum and pallidum in limbic-related ones. Here, the projections of the globus pallidus to the rostral part of the ventrointermediate area (VIAr), similar to VA and VLo in mammals, are shown, as are the pallidal projections to the intralaminar-like DTZ, which were noted above. Like the motor-related nuclei of mammals, VIAr receives inputs from the GABAergic neurons of the substantia nigra pars reticulata and the deep cerebellar nuclei, and it projects to motor-related areas of the Wulst.

Also as in mammals, the dopaminergic inputs of the substantia nigra pars compacta to the striatum are shown, as are the interconnections of the globus pallidus with the subthalamic nucleus. Thus both the so-called direct loop of pallial regions to dorsal striatum to globus pallidus to dorsal thalamus and back to pallium, a circuit that promotes the initiation of movement, and the so-called indirect loop, which involves the subthalamic nucleus and inhibits movement, are present in birds. These circuits are the TPSpT and TPPSpT types, and are comparable to those of mammals. Whereas the mammalian dorsal striatum has a heterogeneous structure of matrix and striosomes, in birds, as with the globus pallidus, the different elements are present but are not anatomically segregated. Birds have additional circuits of the TPPSpT type that involve multiple pallial areas, such as those through the prefrontal cortex-posited homologue candidates of the nidopallium caudolaterale and the dorsolateral corticoid area. Prefrontal-like behavioral deficits in birds, particularly involving working memory paradigms, have been found following lesions that involve these regions (Hartmann and Güntürkün, 1998; Lissek et al., 2002).

In songbirds and parrots, which exhibit complexity of vocalization and communicative abilities, similar forebrain circuit loops involve the song and vocalization nuclei specific to these taxa (Wild, 1993, 1994; Vates et al., 1996; Striedter and Vu, 1998; Bottjer et al., 2000; Jarvis and Mello, 2000; Jarvis et al., 2000, 2002; Lavenex, 2000; Deng et al., 2001). Some of the connections of the auditory-vocal system in songbirds are shown (in purple) in Figure 5, which is organized in the same way as Figures 3 and 4. Incoming auditory and other sensory information is relayed through the collothalamic nuclei ovoidalis and uvaeformis (shown in blue) to parts of the nidopallium within the dorsal ventricular ridge—the nucleus interfacialis and Field L and its subdivisions. From there, the sensory input is relayed to a region, also within the neostriatum, called the higher vocal center, or HVC. Recurrent loops then involve projections from HVC to a specialized part of the striatum called area X, which projects directly to the dorsal thalamic zone, or DTZ. The DTZ in turn projects to two additional sites in the anterior part of the nidopallium, the medial and lateral magnocellular nuclei, which then project to HVC or directly back to area X, respectively. Like the loops discussed above, these song circuit loops thus involve glutamatergic, excitatory components with spontaneous activity as well as a regulatory, inhibitory GABAergic component supplied by area X. The motor output pathway is influenced by these circuits, since it arises in HVC via projections to another part of the dorsal ventricular ridge called the robust nucleus of the arcopallium. From there, descending projections are targeted upon the brainstem, particularly on a division of the hypoglossal nucleus (nXIIts), which innervates the vocal organs.

View larger version (35K): [in this window] [in a new window]   Figure 5. Diagram of the vocal circuits in birds, based on songbirds (Wild, 1993, 1994; Vates et al., 1996; Bottjer et al., 2000; Deng et al., 2001; Jarvis et al., 2002).

	Reptilian and Amphibian Behavioral Abilities and Brain Organization

TOP Abstract Introduction The Complex Cognitive Abilities... Cognition, Consciousness, and... Similarities and Differences in... Reptilian and Amphibian... Conclusions Appendix Glossary of Terms Pertaining... Literature Cited

Behavioral studies on reptiles and amphibians that address cognitive issues are few indeed in comparison to those in birds and mammals. Most comparative studies of cognition focus on birds and mammals and, occasionally, on arthropods or octopuses (Kesner and Olton, 1990; Bekoff et al., 2002). Reptiles can be trained to successfully carry out some behavioral tasks, such as in a simple visual discrimination paradigm (Bass et al., 1973; Bass, 1977), but paradigms to test higher-level cognitive challenges are difficult if not impossible to apply. A literature search for amphibian behaviors in the cognitive range did not yield any results. Thus, the negative statement that these taxa do not exhibit complex cognitive abilities, such as working memory, cannot at present be substantiated by literature citations.

In general terms and excluding the song system nuclei, reptiles have many of the same neural components as many species of birds (Butler and Hodos, 2005). However, many of the circuits are less elaborate, and some circuits are absent or, at best, very sparse. While techniques and the number of studies vary for some of the available information, a consistent picture across systems is discernible. For example, while the basic ascending sensory pathways for the visual, auditory, and somatosensory systems are present (Hall and Ebner, 1970; Foster and Hall, 1978; Ebbesson and Goodman, 1981; Hoogland, 1981; Bruce and Butler, 1984a,b; Kenigfest et al., 1997; Zhu et al., 2005; and see Butler and Hodos, 2005), they are less elaborate than the homologous pathways in birds and mammals; i.e., they involve fewer cell groups or subdivisions of cell groups in the thalamus and pallium.

Both similarities and differences exist in the organization of the dorsal thalamus. As in birds but in contrast to mammals, reptiles have few reciprocal projections from the pallium to dorsal thalamic sensory relay nuclei. Only in turtles (Ulinski, 1986) have reciprocal projections from the visual pallium to the dorsal lateral geniculate nucleus been found, and such projections have not been described in other sensory systems.

The circuitry of the thalamic reticular nucleus also may vary across reptiles. In a crocodile, Pritz and Stritzel (1990) identified a population of thalamopetally projecting neurons, but it contains only a few scattered, apparently GABAergic neurons (positive for glutamic acid decarboxylase). In contrast, turtles have a thalamic reticular nucleus that contains inhibitory GABAergic neurons, receives projections from the pallium, and is reciprocally connected with dorsal thalamic sensory relay nuclei (Kenigfest et al., 2005). The circuitry for reciprocal thalamic inhibition and generation of synchrony that is present in mammals and indicated in birds, as discussed above, thus appears to be present in turtles if not all reptiles. Additionally, turtles have GABAergic pallial interneurons (Reiner, 1991; Colombe et al., 2004). Likewise, basal ganglia circuitry for additional inhibitory regulation of thalamic activity is present in reptiles (Reiner et al., 1998). The differences in these systems in reptiles in comparison to birds and mammals may be ones of degree of elaboration rather than presence or absence.

In reptiles, in contrast to both birds and mammals, the telencephalic pallium, like the thalamus, is less elaborate in both its cytoarchitecture and number of cell groups. For example, in reptiles, the visual system target area of nucleus rotundus (Lohman and van Woerden-Verkley, 1978; Balaban and Ulinski, 1981) apparently does not have subdivisions like the entopallium and perientopallium in birds (Kreutzfeldt and Wild, 2005). Likewise, in the auditory system, only one auditory target area appears to be present within the dorsal ventricular ridge in a lizard (Foster and Hall, 1978), and while two auditory areas have been identified in crocodiles (Pritz and Stritzel, 1992), they do not rival the multiple pallial regions that comprise Field L in birds (Carr and Code, 2000; Reiner et al., 2005). The same comparison holds for multimodal, association pallial areas, which, while present in both the anterior (Andreu et al., 1996) and the caudal part (Lanuza et al., 2002; Novejarque et al., 2004) of the dorsal ventricular ridge in reptiles, are substantially less elaborated than comparable areas in birds, some of which may be involved in working memory functions (Güntürkün and Durstewitz, 2001; Csillag and Montagnese, 2005; Rose and Colombo, 2005).

In addition to the degree of elaboration of some cell groups in terms of number of nuclei, subareas, or both, data on brain-body ratios are of note. The brain-body ratios of both birds and mammals substantially exceed those of reptiles and amphibians (Jerison, 2001). The brain-body ratios of most birds overlap those of mammals, with some birds, such as the corvids, overlapping the lower range for primates (Jerison, 2001; and see Nieuwenhuys et al., 1998). Further, just as in mammals, the large brain-body ratios in birds are in large part due to expansion of the forebrain, and specifically, the pallial association areas of the telencephalon—the nidopallium and mesopallium—which are most expanded in birds with high cognitive abilities, the psittacines (parrots) and passerines (which include corvids) (Lefebvre et al., 2004; Iwaniuk and Hurd, 2005).

Forebrain organization in amphibians is markedly different from that in reptiles, birds, and mammals, with the telencephalon characterized by having only a very small nonlimbic pallium situated between the medial (hippocampal) and lateral (olfactory) pallia (see Butler and Hodos, 2005). The medial pallium is the main ascending target of projections from the rostral part of the dorsal thalamus (Northcutt and Kicliter, 1980), which does not receive direct retinal input but rather inhibitory input from ventral thalamic, retinorecipient cell groups (Roth et al., 2003). Also, in marked contrast to amniotes, in amphibians the more caudal part of the dorsal thalamus, the collothalamus, projects mainly to the striatum and only very sparsely to the lateral-most part of the pallium (Northcutt and Kicliter, 1980). Amphibians appear to have both striatal and pallidal components within their subpallium (Marín et al., 1998; Endepols et al., 2004), but their marked lack of thalamopallial circuitry clearly differentiates them from amniotes in terms of forebrain organization.

In a discussion of the evolutionary origins of consciousness, Sjölander (1997) considers the issue of intermodal binding, the "merging of the senses" (Stein and Meredith, 1993). He posits that the capacity to form an internal representation of the world in which the inputs from the various sensory systems are intermodally integrated represents an evolutionary breakthrough. He argues that reptiles, such as snakes in his example, lack this capacity, but that birds and mammals behave in such a way as to indicate that they do form such representations. Reptiles have intermodal, mapped representations at the neural level of their midbrain roof (Stein and Meredith, 1993), and as noted above, although they have multimodal pallial areas (Lanuza et al., 2002; Novejarque et al., 2004), these are relatively small. Thus, higher-level consciousness and cognitive abilities may require an elaborated pallial capacity, such as is present in both birds and mammals, for such intermodal integration.

	Conclusions

TOP Abstract Introduction The Complex Cognitive Abilities... Cognition, Consciousness, and... Similarities and Differences in... Reptilian and Amphibian... Conclusions Appendix Glossary of Terms Pertaining... Literature Cited

 
Birds clearly have forebrain circuitry that, at least in some respects, is very similar to what is found in mammals. They have all the basic components and many of the pathways that are present in mammals. For comparison across tetrapod taxa, a number of selected neural features are listed in Table 1, along with possible consciousness levels and cognitive abilities.

As indicated in Table 1, many of the pallial differences between reptiles and birds are more those of degree in the development of areas than ones involving major shifts of connections. The Wulst and dorsal ventricular ridge regions of birds are substantially expanded in comparison with the corresponding regions in reptiles. This difference is reflected in the jump in brain-body ratios between reptiles and birds (Jerison, 2001; Nieuwenhuys et al., 1998). The brain-body ratios of amphibians are even less than those in reptiles (Jerison, 2001), and as discussed above, their forebrains lack the thalamopallial circuitry of amniote brains. Although consciousness, even of a higher level, cannot be ruled out in reptiles or amphibians, if such thalamopallial circuitry is crucial to its generation in birds and mammals, it is possible that reptiles have consciousness to a substantially lesser degree, and amphibians may or may not have alternate neural systems to support it.

The theory of consciousness that has served as a guide in our investigation (Cotterill, 1997, 1998, 2001) sees this phenomenon as arising when two feedback routes function in parallel. One of these routes passes through the animal’s surroundings, providing sensory input that is a consequence of the animal’s own muscular movements. That route is in operation even for animals that do not possess the capacity for consciousness. The second route, which is the one vital to consciousness, is purely internal, and it is provided by the loops discussed at length above. When the animal’s (internal) drive mechanism sets up a pattern of muscular movements, by which the creature will explore its surroundings, efference copy signals are dispatched around the internal loops, and these impinge (internally) upon the sensory receptors a fraction of a second before the externally emanating feedback signals arrive from the surroundings. This enables the nervous system to guide the animal’s movements by what is essentially a sophisticated servo mechanism in which the significance of impending deviations between the intended and actual feedback is evaluated by the amygdala and corrected for by the basal ganglia. Each planned sequence of muscular movements represents a question (or series of questions) that the nervous system will put to the surroundings, and it is the need to match the question with the anticipated answer that necessitates a mechanism of attention. It is for this reason that the brain components serving attention, and particularly the nucleus reticularis thalami, act at a position rather early in the efference copy route. In neural terms, the nucleus reticularis thalami is indeed rather close to the premotor cortex, where the intended sequence of muscular movements is set up.

From what has been discussed, it is clear that the circuitry required for this proposed higher-level consciousness mechanism is indeed present in both avian and mammalian brains. Shared neural circuits do not, in and of themselves, reveal whether birds are conscious, of course. The behavioral evidence for their higher cognitive abilities strongly suggests that they are. This evidence includes working memory (Diekamp et al., 2002), transitive inference (Peake et al., 2002; Paz-y-Miño et al., 2004), tool making (Weir et al., 2002; Hunt and Gray, 2003; Kenward et al., 2005), multistability of ambiguous figures (Vetter et al., 2000), episodic memory (Clayton and Dickenson, 1998, 1999; Clayton et al., 2003; Zentall et al., 2001), object constancy (Dumas and Wilkie, 1995; Pollok et al., 2000; Pepperberg et al., 1997), verbal and numerical abilities (Pepperberg, 1999; Pepperberg and Gordon, 2005), and theory of mind (Emery and Clayton, 2001; Clayton et al., 2003; Bugnyar and Kotrschal, 2002, 2004). If this view proves tenable, it would indicate that the brain features shared by birds and mammals are sufficient to produce higher-level consciousness, whereas the brain features that are not shared are not essential to the phenomenon.

A promising way to test this hypothesis may derive from the examination of the brain circuitry of very distantly related animals—some of the invertebrates. As Greenspan and van Swinderen (2004) recently have discussed, some arthropods, including the fruit fly Drosophila, honey bees, and the jumping spider Portia, exhibit behavioral and neural responses that indicate explicit memory, attentiveness to stimuli, or both. They are able to use map-like, spatial memory to navigate and also to exhibit spatial forward planning and anticipation. Highly complex, large neuropil regions in the rostralmost part of the brain in these animals are implicated in these cognitive functions. Comparisons of the circuitry of these neuropils with that present in the forebrains of both birds and mammals may allow identification of the crucial components of this circuitry that underlie and support complex cognition. Whether these same neural components likewise support consciousness in any invertebrate species remains to be considered.

	Appendix

TOP Abstract Introduction The Complex Cognitive Abilities... Cognition, Consciousness, and... Similarities and Differences in... Reptilian and Amphibian... Conclusions Appendix Glossary of Terms Pertaining... Literature Cited

 
Overview of Forebrain Organization in Amniotes
This section presents a brief survey of the forebrain structures in mammals and birds (Nieuwenhuys et al., 1998; Butler and Hodos, 2005) that are of particular relevance to the circuitry discussed in the present paper. In all amniotes, the telencephalon has a dorsal (upper) part, the pallium (shown in light shading in Fig. 2), and a ventral (lower) part, the subpallium (shown in darker shading). In mammals (Fig. 2A), the pallium includes a hippocampal part involved in learning and memory; an olfactory-recipient region, part of which is called piriform cortex; and the very extensive region of neocortex. The neocortex, with its unique cytoarchitecture of six cell and fiber layers, is the hallmark of the mammalian brain and the site where sensory and motor processing are mediated and where, in concert with thalamic loop systems, consciousness and thought are generated. In the most lateral part of the pallium lie two deeper formations, the claustrum-endopiriform formation and the pallial amygdala, which are both elaborations of the deeper layers of the pallial mantle. The lateral nucleus of the amygdala, in particular, receives ascending sensory projections. The subpallium, in contrast, contains various structures, including the septal nuclei, a subpallial part of the amygdala, and the basal ganglia. The latter comprise a dorsal pair of striatal (caudate nucleus and putamen) and pallidal (globus pallidus) components (shown in Fig. 2A) and a ventral striatopallidal complex as well.

In birds (Fig. 2B), the pallium includes a hippocampal part, similar to the hippocampus of mammals in many if not all respects. It also has an olfactory-recipient piriform cortex that lies laterally (not present at the level shown in Fig. 2B). Rather than having neocortex, much of the pallial area comprises the Wulst, or hyperpallium, which has several divisions, and the dorsal ventricular ridge, which includes the mesopallium and nidopallium, as shown in Figure 2B. Additionally, the entopallium is a visual part of the nidopallium. As in mammals, birds have a subpallium with multiple components, including striatal and pallidal parts that exhibit multiple similarities with their counterparts in mammals. Some pallial amygdalar components lie more caudally within the dorsal ventricular ridge, as do an area called arcopallium and a subpallial amygdalar area, but a homologue of the sensory-recipient lateral amygdala of mammals has not been identified with certainty.