I'm teaching a course in neurobiology this term, and it's strange how it warps my brain; suddenly I find myself reaching more and more for papers on the nervous system in my reading. It's not about just keeping up with the subjects I have to present in lectures (although there is that, too), but also with unconsciously gravitating toward the subject in my casual reading, too.

"Unconsciously"…which brings up the question of exactly what consciousness is. One of the papers I put on the pile on my desk was on exactly that subject: Evolution of the neural basis of consciousness: a bird-mammal comparison. I finally got to sit down and read it carefully this afternoon, and although it is an interesting paper and well worth the time, it doesn't come anywhere near answering the question implied in the title. It is a useful general review of neuroanatomical theories of consciousness—even if it left me feeling they are all full of crap—but in particular it's an interesting comparative look at avian brain organization.

The paper briefly reviews four classes of models that attempt to locate the centers of consciousness, or "consciousness generators", in the mammalian brain. Alas, they all seem to contradict one another, and are all driven by the authors' hypotheses to invent validation in the structure, rather than by actual data that might lead to useful hypotheses. I won't get into these, other than to paste in the summary—while I'm not at all dazzled by any of them, it is handy to have such tidy summaries.

Classification of consciousness—brain theories

	Bottom-up	Top-down
Sensory systems	A	B
Motor systems	C	D

A: A representative bottom-up theory proposed by Crick and Koch concerns the visual system and asserts that visual awareness is associated with activity in higher order visual areas that are in direct contact with prefrontal cortex. Although the "cortical system" covered by this theory includes the entire cerebral cortex, dorsal thalamus, claustrum (a nonlaminated structure deep to the cortex), and dorsal striatopallidal complex (caudate, putamen, globus pallidus—also referred to as basal ganglia and involved in motor control) in the forebrain as well as the motor control-related cerebellum and various brainstem projection systems, the generator neurons seem to be limited to temporal, parietal and prefrontal regions of the neocortex. Crick and Koch limit the generator structure further by assuming that activity in a subpopulation of neurons in cortical layer V, characterized by firing in burst patterns, is crucial. A prominent feature in the theory of Crick and Koch is the insistence that the primary visual cortex is not a generator structure.

B: The theory of Edelman and Tononi appears to be an example of a top-down sensory approach and focuses on the general features of consciousness—such as complexity and unity. It asser ts that consciousness is associated with activity in the temporal and frontal associative and motor regions of the cortex: a "dynamic core", characterized by "re-entrant" interactions within limited portions of the CNS. Possibly, thalamic neurons are included in the dynamic core. Structures supporting the dynamic core seem to be the septal region, amygdala, hippocampus, dorsal thalamus, and hypothalamus within the forebrain, as well as the reticular activating system in the brainstem. Edelman and Tononi seem to assume a larger and more dynamic population of generator neurons than Crick and Koch. In comparison to the latter, the inclusion of limbic system structures—septal region, amygdala, and hippocampus—that are related to emotion and learning make this theory more broadly based.

C: The theory suggested by Eccles may be said to represent a bottom-up approach, largely based on studies of the motor system. It takes consciousness to be associated with activity in cortical columns of the pre- or supplementary motor areas, for med by groups of generator pyramidal cells, organized in a specific way (dendrons). This means that Eccles's theory assumes the smallest and most well defined population of generator neurons of all the theories discussed here.

D: The theory formulated by Cotterill may be said to be an example of a motor system top-down approach. According to this, consciousness is associated with activity in a circuit consisting of sensory and motor, cortical and thalamic structures. Fast feedback from muscle activity, making muscle spindles (sensory receptors for the degree of stretch within muscles) critical for the generation of consciousness, is a central par t of the theory. Fundamental to Cotterill's, as well as to other top-down theories, is a set of proposals implying that the neural basis for consciousness lies in the ability of an organism to know itself, or its proto-self, within its environment, by bodily movements and homeostatic functions. In all these top-down theories, several non-cortical regions seem to be included in the generator structure of consciousness. Cotterill includes the amygdala, hippocampus, dorsal thalamus, subthalamus, hypothalamus, and dorsal striato-pallidal complex (caudate, putamen, and globus pallidus) within the forebrain as well as multiple brainstem structures—the superior colliculus (involved in sensory and motor mapping of space), various structures involved with motor system regulation (cerebellum, substantia nigra, pontine nucleus, red nucleus, and inferior olive), and part of the autonomic nervous system (for control of visceral functions)—thus for ming a larger generator structure than any other theory discussed here. Additional motor system top-down ap- proaches include that of John, who includes the thalamus, limbic system, and dorsal striato-pallidal complex, and that of Parvezi and Damasio,(24) who incorporate the hypothalamus, intralaminar and reticular nuclei of the dorsal thalamus, basal forebrain, and various cholinergic, glutamatergic, noradrenergic, dopaminergic and serotonergic projection systems that regulate the activity of the cortex.

There's also a diagram of where these theories place the "conciousness generators" (in dark gray) in the human brain. As you can see, they all agree it is not in the brainstem or cerebellum, but from there anything goes.

Diagrammatic representations of the distribution of generator neurons (shaded in gray) for the four principal groups of brain-consciousness theories outlined in the present essay as they would appear for the human brain. For each theory, the lateral aspect of the human brain is shown on the left. The entire extent of the frontal lobe is shaded in B at the rostral (to the right) end of the cerebral hemisphere; within it lie motor cortices and, rostral to them, the prefrontal cortex. The occipital lobe is unshaded in both A and B, lying at the caudal pole (to the left) of the hemipshere. The parietal lobe (dorsally) and temporal lobe (ventrally) lie between the occipital and frontal lobes. The medial aspect of the hemisphere is shown for each theory on the right. The hippocampal formation, olfactory (piriform) cortex, and amygdala all lie within the temporal lobe, deep to the neocortical areas shown here.

The more productive part of the paper is the comparison between mammals and birds. Here's the premise:

We posit that, since highly complex cognitive abilities are correlated with presumed consciousness in at least some mammals, including but not limited to humans, and since highly complex cognitive abilities are evinced by birds, it is likely that consciousness is also present in birds. Given that hypothesis, we then can compare the anatomical organization of mammalian and avian brains. We reason that if (1) complex cognition and consciousness are present in both mammals and birds and (2) consciousness has any neural basis, then birds should have at least some neural features in common with mammals to generate consciousness.

That sounds promising, but several problems come to mind. 1) If no one can even agree on the neural features responsible for consciousness within mammals, how is this comparison going to identify commonalities? 2) Birds and mammals are related lineages, so many brain similarities are going to be consequences of shared history, not function. Why not go all out and compare more distinct lineages…say cats and octopus? 3) Since we don't even know what features of the areas of the brain are responsible for consciousness, we aren't going to be able to recognize if different regions in birds and mammals have independently acquired whatever mysterious property is involved. While I think the comparative approach is terrific, in this case it's premature and targeted at the wrong level.

But hey, you've got to start somewhere, and this wasn't one of those exasperating papers that I toss into the wastebasket. It has a good summary of the evidence for avian intelligence, listing the various features they've exhibited.

• Transitive inference. You can train pigeons (Pigeons! Birds that are archetypically stupid!) to recognize rank order, such as that A<>B, and B<C, and they can use that information to recognize tha A<C.
• Coherence. Pigeons can respond variably to the ambiguity in figures like the Necker cube. That suggests that they have a mental model of what it should be, and their impression of its orientation can "flip".
• Episodic memory. Scrub jays can recollect when, where, and what is stored in their food caches.
• Piagetian object constancy. Doves, magpies, parrots, and ravens aren't fooled if an object is hidden—despite being out of sight, they have a mental model of its position.
• Cognitive abilities. Gray parrots are singled out as particularly brilliant, with individuals able to count to 7, recognize the concept of "none", and able to understand the concepts of "same" and "different".
• Tool use. Ravens, parrots, and New Caledonian crows have all been shown to be able to make and use simple tools.
• Theory of mind. Scrub jays are able to attribute their own predilections to other members of the same species. Jays that rob caches are more likely to move their own caches than "honest" jays.

While I have my doubts about the neuroanatomical comparisons, the authors bring up one very general point. In mammals, the neocortex—the hugely enlarged part of our forebrains that is the first thing you see when you crack open our skulls—is central to higher level thinking. The comparable structure in birds is called the hyperpallium, or Wulst (German for "bulge"), and the posterolateral portion of it is an important visual center, comparable to the striate cortex, or visual areas of our brain. The fascinating thing is that the cellular organization of these two areas with similar functions and perhaps similar roles in generating consciousness are very different.

The mammalian visual cortex is characterized by a beautifully layered organization. Different inputs segregate to different layers, and pyramidal neurons extend long dendrites orthogonal to those layers, like long antennae reaching up and sampling incoming data streams, each of which is segregated spatially. The Wulst, on the other hand, is organized like other nuclei of the brain, and the primary neurons are star-shaped, reaching out in all directions to contact their inputs. They lack the rigid but elegantly arrayed spatial segregation seen in us mammals.

The pictures below don't really do it justice, but a good neurohistologist (sometimes even a barely adequate one, like me) can pick out 6 discrete layers in an appropriate slice of mammalian visual cortex. The bird Wulst is alien-looking. Huge, but weird.

Photomicrographs of the visual lemnopallium in the barn owl (Tyto alba), common ferret (Mustela putorious) and eastern tube-nosed megabat (Nyctimene robinsoni). In the owl, the photomicrograph depicts a coronal section stained for Nissl substance (localized in cell bodies) through the visual hyperpallium, the Wulst (pial surface to the top of the page). Note the thick outer layer termed the hyperpallium apicale (HA), the high cell density of the nucleus interstitialis hyperpallii apicalis (IHA), and the deepest portion termed the hyperpallium densocellulare (HD). This region of the Wulst, forms what are termed pseudolayers, which are best thought of as flattened and stretched out nuclei rather than true layers as is evident in the cerebral cortex of the ferret and megabat pictured here. This architecture contrasts with the typical 6-layered cerebral cortex found in the primary visual cortex of mammals.

In their conclusion, the authors are a bit vague about the relevance of the earlier theories of consciousness to bird neuroanatomy. Parts fit, others don't, but since I think the theories are so nebulous that it's nearly impossible to draw any conclusions from that, they can't come down strongly one way or another. One suggestive observation is that bird brains are more similar to reptilian brains than mammalian brains are to stem amniotes'. If birds are conscious, that makes the assumption that the capacity for consciousness arose at that stem amniote-mammalian border suspect. The capacity, in the sense of having neural circuitry that could be adapted to generate consciousness, could have been present earlier.

Ann B. Butler, Paul R. Manger, B.I.B. Lindahl, Peter Århem. Evolution of the neural basis of consciousness: a bird-mammal comparison (p 923-936).