Animal consciousness: a synthetic approach.

Abstract

Despite anecdotal evidence suggesting conscious states in a variety of non-human animals, no systematic neuroscientific investigation of animal consciousness has yet been undertaken. We set forth a framework for such an investigation that incorporates integration of data from neuroanatomy, neurophysiology, and behavioral studies, uses evidence from humans as a benchmark, and recognizes the critical role of explicit verbal report of conscious experiences in human studies. We illustrate our framework with reference to two subphyla: one relatively near to mammals - birds - and one quite far -cephalopod molluscs. Consistent with the possibility of conscious states, both subphyla exhibit complex behavior and possess sophisticated nervous systems. Their further investigation may reveal common phyletic conditions and neural substrates underlying the emergence of animal consciousness.

Full text

Page 1

Animal Consciousness: A Synthetic Approach
1
2
David B. Edelman1 & Anil Seth2 3
1The Neurosciences Institute, 10640 John Jay Hopkins Drive, 4
San Diego, CA 92121, USA 5
david_edelman@nsi.edu 6
7
2Department of Informatics, University of Sussex, Brighton BN1 9QJ, UK 8
a.k.seth@sussex.ac.uk 9
10
Abstract 11
Despite anecdotal evidence suggesting conscious states in a variety of non-human 12
animals, no systematic neuroscientific investigation of animal consciousness has yet been 13
undertaken. We set forth a framework for such an investigation that incorporates 14
integration of data from neuroanatomy, neurophysiology, and behavioral studies, uses 15
evidence from humans as a benchmark, and recognizes the critical role of explicit verbal 16
report of conscious experiences in human studies. We illustrate our framework with 17
reference to two subphyla: one relatively near to mammals—birds—and one quite far—18
cephalopod molluscs. Consistent with the possibility of conscious states, both subphyla 19
exhibit complex behavior and possess sophisticated nervous systems. Their further 20
investigation may reveal common phyletic conditions and neural substrates underlying 21
the emergence of animal consciousness. 22
23
(terms in bold appear in the glossary) 24
25
26

Page 2

A synthetic framework for studying animal consciousness. 26
27
Although Darwin proposed that animal and human minds alike are the products of
natural selection 1, questions of animal consciousness were largely neglected throughout
the 20th century (but see Griffin2; note the term ‘animal’ is used here to mean ‘non-human
28
29
30
animal’). This neglect may have arisen in part because, seemingly, only humans are 31
capable of accurately describing their phenomenal experience. However, there is now 32
abundant and increasing behavioral and neurophysiological evidence consistent with, and 33
even suggestive of, conscious states in some animals. We will use humans as a 34
benchmark for the development of new empirical criteria for further investigation. Here, 35
we apply this approach to birds and cephalopod molluscs, subphyla that exhibit complex 36
cognitive faculties and behaviors and have strikingly elaborate brains. These two 37
subphyla are examples of highly distinct lineages, and their study provides an excellent 38
opportunity to examine how conscious states might be instantiated in very different 39
nervous systems. While we do not resolve this issue here, we propose that its 40
examination lies within the reach of contemporary neuroscience. 41
42
Humans as a benchmark. The notion that consciousness can be engendered in 43
different nervous systems by a variety of underlying mechanisms suggests a need to 44
examine constraints, and therefore to synthesize behavioral, neurophysiological, and 45
neuroanatomical evidence. Human studies involving the correlation of accurate report 46
with neural correlates can provide a valuable benchmark for assessing evidence from 47
studies of animal behavior and neurophysiology. A constraint on this strategy is that the 48
capacity for accurate report of conscious contents implies the presence of higher-order 49
consciousness, which in advanced forms may require linguistically-based narrative 50
capability. This is in contrast to primary consciousness, which entails the ability to
create a scene in the ‘remembered present’3 in the absence of language. Primary
51
52
consciousness may be a basic biological process in both humans and animals lacking true 53
language. 54
Various properties of human consciousness can be identified at neural, behavioral,
and phenomenal levels 4. Neural correlates of human consciousness include the presence
55
56

Page 3

of thalamocortical signaling, fast, irregular, low-amplitude electroencephalographic
(EEG) signals, and widespread cortical activity correlated with conscious contents 5-7. At
the behavioral level, consciousness has been associated with behavioral flexibility 8,
rational action 9, and certain forms of conditioning 10. These can be related to cognitive
properties involving widespread access and associativity 8, multiple discriminations 11,
and the capacity for accurate report 5. These properties can be mapped to a variety of
functions related to consciousness 12. At the phenomenal level, human consciousness
57
58
59
60
61
62
63
involves the presence of a sensorimotor scene, the existence of a first-person perspective,
the experience of emotions, moods, and a sense of agency 13 14.
64
65
Using humans as a benchmark, behavioral, cognitive, and neural properties can be 66
employed as empirical criteria informing the ascription of conscious states to animals 67
(Figure 1). The application of this approach requires that: (i) at least some of these 68
properties reliably occur in both humans and animals; (ii) human brain areas responsible 69
for consciousness can be seen to be integrated with those areas responsible for accurate 70
report of phenomenal experience; and (iii) neural evidence that can be correlated with 71
phenomenal properties of consciousness must in addition account for those properties. 72
73
In humans, explicit verbal, or linguistic, report of a conscious experience is 74
sometimes taken as a ‘gold standard’ in the sense that it guarantees the presence of 75
consciousness. However, many creatures, including infants, most animals, and aphasic 76
human adults, are constitutively unable to produce linguistic reports. The production of 77
such reports is therefore too limiting a criterion for ascription of consciousness in general. 78
Importantly, accurate report may exploit behavioral channels other than language, for 79
example lever presses or eye blinks (Box 1). However the ability to provide such report 80
nonetheless implies the presence of higher-order (metacognitive) access to primary 81
conscious contents, which may not be constitutively required for primary consciousness. 82
Our approach therefore recognizes that the mechanisms responsible for primary 83
consciousness may be distinct from those mechanisms enabling its report. 84
85

Page 4

The extent to which neural evidence can account for phenomenal properties is 86
particularly important with respect to those properties that are common to most or all 87
conscious experiences. For example, in humans, every conscious scene is both integrated
(i.e., ‘all of a piece’) and differentiated (i.e., composed of many different parts) 11.
88
89
Therefore, finding neural processes that themselves exhibit simultaneous integration and 90
differentiation would help to explain, and not merely correlate with, the corresponding 91
phenomenal property. Such neural processes can therefore be considered to be
‘explanatory correlates of consciousness’ 14 , and because they are explanatory, their
92
93
identification in animals is more suggestive of the presence of corresponding phenomenal 94
properties than is the identification of neural correlates per se. In this view, conscious 95
states are neither identical to neural states nor are they computational or functional 96
accompaniments to such states; rather, conscious states are entailed by neural states in 97
much the same way that the spectroscopic response of hemoglobin is entailed by its
molecular structure 6.
98
99
100
The presence of voluntary behavioral responses is another candidate ‘gold 101
standard’ for the ascription of consciousness. However, the absence of such responses in 102
humans does not necessarily imply absence of phenomenal experience. For example, in a 103
recent study of a patient in a behaviorally unresponsive vegetative state, brain activity
related to volition was taken as persuasive evidence of residual consciousness 15.
104
105
Conversely, behavior that appears to be volitional could be attributed widely among 106
animals on the basis of spontaneous and adaptive behavioral responses. Therefore, 107
apparent voluntary behavior at best provides a weak criterion for the ascription of 108
consciousness. In addition, conditions such as the vegetative state underscore the point 109
that consciousness should not be confused with arousal or inferred directly from the 110
existence of distinct sleep/wake cycles. States resembling deep sleep have been
observed in many animals, including Drosophila melanogaster 16 and Caenorhabditis
elegans 17; conversely, female killer whales and dolphins and their newborn calves may
not sleep for periods of four to six weeks postpartum 18.
111
112
113
114
115

Page 5

Mammalian consciousness: Extending the benchmark. Mammals, particularly 116
primates, share with humans many neurophysiological and behavioral characteristics 117
relevant to consciousness, and therefore represent a relatively uncontroversial case for the
ascription of at least primary consciousness 5. In a classical example, Logothetis et al.
118
119
trained rhesus macaque monkeys to press a lever to report perceived stimuli in a
binocular rivalry paradigm 19 (Box 1). Neurons in macaque inferior temporal (IT)
120
121
cortex showed activity correlated with the reported percept, whereas neurons in the visual 122
area V1, instead responded to the visual signal. This suggests a critical role for IT in 123
visual consciousness. These observations are consistent with evidence from humans 124
subjected to binocular rivalry while being examined via magnetoencephalography 125
(MEG). The results from these studies suggest that consciousness of an object involves
widespread coherent synchronous cortical activity 20. This correspondence between
126
127
monkeys and humans provides an example of how benchmark comparisons across 128
humans and animal species can be made. With this in mind, we now explore the 129
application of benchmark comparisons to two widely divergent animal subphyla: birds 130
and cephalopod molluscs. We are aware of the pitfalls of making facile comparisons and 131
implying that homologies exist in the absence of strong evolutionary evidence. In each 132
case, we will present behavioral evidence first, and then present evidence based on neural 133
architecture and dynamics. 134
135
Building a Case for Avian Consciousness 136

Avian cognition and behavioral capabilities. Feats of avian memory 21, 22, tool use and
manufacture 23 , deception 24, and vocal learning and performance 25, 26, the capacity of
some species to employ lexical terms in meaningful ways 27, and evidence for higher
order discriminations in some birds 27, 28 collectively support the functioning of nervous
137
138
139
140
141
systems as sophisticated as those of most mammals. 142
143
Episodic and working memory capabilities are implied by the sophisticated food
caching behaviors of corvids (e.g., jays, jackdaws, magpies, rooks, crows, and ravens) 23
144
145
and by laboratory-based demonstrations of transitive inference and delayed-match-to-146

Page 6

sample in pigeons and great tits 29. African grey parrots, magpies, and ring doves have 147
shown the ability to track periodically hidden and displaced objects; such object
constancy certainly requires working memory 30. In addition, spatial learning, though
148
149
obviously implied by navigation during flight, has been shown explicitly by hooded
crows learning to negotiate a radial maze 31.
150
151
152
Studies of avian tool use and manufacture imply not only elaborate memory and 153
learning substrates, but also the ability to make sophisticated discriminations and to plan 154
behaviors before executing them. For example, New Caledonian crows have been 155
observed to fashion hooked-wire tools to retrieve food from a glass cylinder, sometimes
flying to a distant perch to bend the wire before returning to the cylinder 23. Similarly,
156
157
wild crows are known to fabricate tools from twigs and leaves in order to extract insects
from holes in trees 23. In addition, a variety of avian species, including Japanese quails32
and European starlings 33, may be capable of social or observational learning (but see
Zentall 34 and Heyes 35). Perhaps most notable are instances in which scrub jays re-
cached food in private after their initial caching was witnessed by conspecifics 30, and
158
159
160
161
162
observations of ravens challenging conspecifics that witnessed caching activities while
ignoring ‘naïve’ birds 36. These behaviors suggest that some birds may be capable of
163
164
theory of mind. 165
166
The avian vocal channel. While the foregoing cognitive capabilities are suggestive of 167
conscious states, the most promising avenue for investigating avian consciousness may 168
involve the study of species capable of vocal learning, which enables a highly flexible 169
form of accurate report. The capacity for some form of vocal learning is shared by at 170
least six animal groups, including cetaceans, bats, parrots, songbirds, hummingbirds,
elephants, and possibly even mice and some other rodents 37. In birds, vocal learning
171
172
enables sophisticated song learning and production, mimicry of sounds, and, in the
psittacines (parrots), word production, comprehension, and naming 27. For example,
173
174
African grey parrots were able to name objects, having acquired vocabularies roughly 175
equivalent to those of some language-trained chimpanzees (albeit after years of training
and reinforcement) 38. Indeed, by naming objects in categorization paradigms, these
176
177

Page 7

animals appeared to produce accurate reports of sophisticated discriminations they were 178
making. ‘Alex,’ a principal subject of many of these experiments, when presented with 179
an altered array of objects, seemed able to make a judgment to the effect that, ‘‘I know 180
that something in this perceptual scene has changed, and here is what has changed.’’ 181
This finding suggests the ability to make discriminations about putative primary
conscious states that appears to resemble some form of higher order consciousness 27 28.
182
183
184
185
Avian neural structures and processes homologous to those of mammals suggest 186
possible neural substrates for both consciousness and its report. 187
188
On what structural bases might avian vocal behavior be related to structures 189
underlying human verbal accurate report? Supporting evidence could come from shared 190
neural mechanisms. One example may be the neural substrate for motor learning in 191
mammals and that for song learning in some birds. Much of the neural basis for song 192
learning in oscines (songbirds) and psittacines resides in an anterior forebrain pathway
involving the basal ganglia, in particular, a striatal nucleus called Area X 39 (see Figure
193
194
2). The anatomical and physiological properties of neurons in Area X closely resemble 195
those of neurons in the mammalian striatum. Specifically, the four neuronal phenotypes 196
found in mammalian striatum are also present in Area X. A notable difference is the 197
presence of a fifth neural phenotype in Area X—but not in mammalian striatum—that is 198
similar to cells found in the mammalian globus pallidus. Area X may therefore comprise 199
a novel mixture of striatal and pallidal anatomies, but it is nonetheless recognizably 200
homologous to the direct striatopallidothalamic pathway of the mammalian basal ganglia 201
39. A similar circuit has been reported in the anterior forebrain of the budgerigar, a parrot 202
40. Together, these findings strongly suggest common functional circuitry underlying the 203
organization and sequencing of motor behaviors related to vocalization in birds and 204
mammals capable of vocal learning. However, whether in some birds this circuitry is 205
embedded within a broader network homologous to that underlying human verbal report 206
remains to be determined. 207
208

Page 8

Another avenue for exploring non-mammalian consciousness is to identify 209
structural and functional homologs to mammalian thalamocortical systems. Vertebrate 210
nervous systems follow a highly conserved body plan that emerged with the first 211
chordates more than 500 million years ago. Consequently, many vertebrate neural 212
structures can be traced to common origins in specific embryological tissues. Avian 213
homologs of subcortical structures, such as the hypothalamus and pre-optic area, are 214
relatively easy to recognize. Although the identity of the avian neural homolog of
mammalian isocortex remains controversial 41, 42, comparative embryological studies
215
216
suggest that the basic underlying neuronal composition and circuitry of the mammalian 217
cortex were established within clustered arrangements of nuclei long before the
appearance of the distinct six-layered mammalian cortex 43. In particular, the nuclei
218
219
comprising the dorsal ventricular ridge (DVR) of the developing avian brain contain 220
neuronal populations homologous to those present in different layers of the mammalian 221
neocortex. These include neurons receiving thalamic input, as well as cells projecting to 222
brainstem and spinal cord neurons. The neurons of the avian DVR and mammalian 223
cortex are nearly identical in both their morphology and constituent physiological
properties 44.
224
225
226
Structural homologies can also be identified using molecular and 227
immunohistological techniques. In particular, neurotransmitters, neuropeptides, and 228
receptors specific to particular neuronal populations within mammalian brain regions 229
have been localized to homologous avian brain regions. For example, both AMPA 230
receptor subunits and the pallidal neuron/striatal interneuron marker Lys8-Asn9-231
neurotensin8-13 (LANT6) are found in the neurons of both mammalian and avian basal
ganglia 45, 46. Finally, gene expression patterns similar to those of mammals have been
232
233
identified in the avian brain. For example, a comparison of homeotic genes involved in 234
early brain development in chick and mouse embryos has revealed robust structural
homologies between parts of the avian telencephalon and mammalian cortex 47.
235
236
237
Deep avian-mammalian homologies have also been revealed by examining 238
functional properties of neuronal populations within particular brain regions. The avian 239

Page 9

anterior forebrain pathway may be functionally analogous to the mammalian corticobasal 240
ganglia–thalamocortical loop. This is suggested by the presence of both inhibitory and 241
excitatory pathways in the medial nucleus of the avian dorsolateral thalamus (DLM), as 242
well as by functional similarities between neurons in the DLM and mammalian 243
thalamocortical neurons. Similarities have also been found among the excitatory and 244
inhibitory circuitry of birds and mammals, particularly in the serotonergic, GABAergic,
and dopaminergic systems 48.
245
246
247
In addition to neuroanatomy, electrophysiological studies are critical in 248
establishing functional homologies between avian and mammalian nervous systems. 249
Currently, however, common properties of mammalian thalamocortical neurons, such as 250
low-threshold calcium (Ca2+) spikes and slow oscillations, have not yet been found in
birds 49. Nonetheless, similarities between the waking EEG patterns of birds and
mammals, as well as slow wave electrical activity recorded during avian sleep 50, are
251
252
253
suggestive of neural dynamics that might support conscious states in birds. 254
255
The existence in birds of structural and functional homologies to mammalian 256
thalamocortical systems is certainly consistent with the presence of higher cognitive
faculties and perhaps consciousness 30, 51. Nevertheless, a compelling case for avian
257
258
consciousness cannot be made solely on the strength of relevant neuroanatomical and 259
neurophysiological resemblances. Nor are descriptions of avian behaviors that imply 260
sophisticated cognitive capabilities sufficient to make such a case. New experimental 261
strategies are needed for evaluating possible conscious discriminations in awake, 262
behaving birds. The findings obtained from studies of Alex the African grey parrot 263
encourage the development of such strategies. Even more challenging though are 264
approaches to investigating conscious behavior in invertebrates. 265
266
Searching for Consciouness in Invertebrates: The Cephalopod Case 267
268
The richness of cephalopod behavioral repertoires. In addition to possessing large 269
brains, cephalopod molluscs have extremely flexible behavior and highly developed 270

Page 10

attentional and memory capacities that may be suggestive of conscious states 52. The 271
performance of some cephalopods (particularly Octopus vulgaris) in several learning and
memory paradigms (e.g., flexibility, persistence of memory traces, contextual learning)53
272
273
is formidable, and comparable in sophistication to that of some vertebrates. Octopuses
can make discriminations between different objects based on size, shape, and intensity 54,
274
275
55, classifying differently shaped objects in the same manner as vertebrates ranging from
goldfish to rats 55. Octopuses are also capable of finding the correct path to a reward in a
plexiglas maze and can retrieve objects from a clear bottle sealed with a plug 56, 57. In
276
277
278
another striking study, ‘naïve,’ or ‘observer,’ octopuses watched conditioned animals 279
(‘demonstrators’) choose between two simultaneously presented objects that differed in 280
contrast only; the observer octopuses later made the same contrast choices in isolation
and without any explicit conditioning 58. Although controversial 59-62, this finding
281
282
suggests that octopuses are capable of observational learning, a faculty previously 283
thought to be unique to highly social animals. 284
285

both the octopus and the cuttlefish 53, 63. In a maze containing obstacles that were
Finally, distinct capacities for short- and long-term memory have been shown in 286
287
changed ad libitum; octopuses could remember these changes and adjust their movements 288
accordingly. Interestingly, the octopuses in this study appeared to pause and deliberate
about the layout of the maze before proceeding 64.
289
290
291
Cephalopod brains: complex nervous systems distant from the vertebrate line. The 292
organization of invertebrate nervous systems diverges so greatly from those of 293
vertebrates such as birds and mammals that, until recently, sophisticated cognitive
capabilities had rarely been ascribed to invertebrate species (see Figure 2). Bees 65, 66,
spiders 67 68, and the cephalopod molluscs 52 are notable exceptions.
294
295
296
297
Of all cephalopod molluscs, the octopus has the largest population of sensory 298
receptors. These receptors communicate with a nervous system that, in adults, may
contain between 170 million and 500 million cells, most of which are neurons 69, 70. In
299
300
the brain of one genus of squid, Loligo, at least 30 distinct nucleus-like lobes have been 301

Page 11

identified 71. The optic lobe, the largest of the fused central ganglia, contains as many as 302
65 million neurons. In addition to processing visual input, the optic lobe plays a critical
role in higher motor control and the establishment of memory 69. A number of other
lobes may be functionally equivalent to vertebrate forebrain structures 69, though their
303
304
305
organization bears little resemblance to the laminar sheets of mammalian cortex. In 306
particular, the vertical, superior frontal, and inferior frontal lobes of octopus, squid, and
cuttlefish are involved in memory consolidation 63, 72, 73. In experiments in which the
307
308
vertical lobe of Octopus vulgaris was lesioned, the ability to learn visual discriminations
was severely impaired, but long-term memory consolidation remained intact 63. Removal
of the median inferior frontal lobe caused memory deficits that compromised learning 54,
309
310
311
74. Taken together, these studies suggest that some regions of the octopus frontal and
vertical lobes are functionally comparable to regions of mammalian cortex (see Young 75
312
313
for a review). 314
315
Radical differences between cephalopod nervous systems and those of vertebrates 316
are exemplified by the parallel, distributed architecture of the octopus locomotor system. 317
The number of neurons in the tentacles of the octopus collectively exceeds the total
number in the central fused ganglia of the brain itself 70. A detached octopus arm will
flail in a realistic manner when stimulated with short electrical pulses 76, suggesting
318
319
320
pseudoautonomous control of some locomotor behavior patterns and hinting at a 321
sophistication of sensorimotor coordination rivaling that of many vertebrates. This 322
elaborate bodily representation in the service of sensorimotor coordination for adaptive
behavior (e.g., locomotion, camouflage, mimicry) might support a ‘core selfhood’ 77 (see
323
324
Box 2), a tantalizing concept as applied to cephalopods. 325
326
With regard to neuropharmacological and physiological properties, the 327
cephalopod nervous system contains many of the major neurotransmitters that are found
in mammalian brains69, 78. In particular, the presence of dopamine (DA), noradrenaline
328
329
(NA), and serotonin (5-HT) receptor subtypes that resemble those found in vertebrates 330
may reflect the presence of circuitry similar to vertebrate excitatory and inhibitory 331
pathways. As is the case with functional avian neuroanatomy, application of 332

Page 12

immunohistochemical and genetic techniques may help determine cephalopod functional 333
analogs to neural regions in mammals that show correlated activity during conscious 334
behavior. An encouraging indication is the recent identification of a cephalopod ortholog 335
to the FoxP2 gene, which in birds and humans has been implicated in motor function 336
related to song and language production, respectively. Notably, FoxP2 expression has
been observed in the adult octopus chromatophore lobes79.
337
338
339
What can be said of neurodynamics in cephalopod brains? Examination of 340
octopus vertical lobe slices has identified long-term potentiation (LTP) of glutamatergic
synaptic field potentials similar to those found in vertebrates 80. More directly related to
341
342
possible conscious states, electrophysiological studies have identified EEG patterns, 343
including event related potentials, which resemble those of awake vertebrates, and at the
same time are distinct from those recorded in other invertebrates.81, 82. Identifying
344
345
cephalopod EEG patterns that reflect low amplitude fast irregular activity similar to that 346
observed during human conscious states will require determination of suitable recording 347
sites. Optic, vertical, and superior lobes of the octopus brain—all of which are critical to 348
learning and memory—are relevant candidates. 349
350
The similarities discussed above by no means confirm the existence of conscious 351
states in cephalopod molluscs, but neither do they exclude them. An intermediate effort 352
to clarify the situation might be the pursuit of psychophysics in cephalopods, an approach 353
not yet represented in the literature (see Box 1). 354
355
Concluding remarks. Approaches to animal consciousness require both clear theoretical 356
frameworks and relevant experimental evidence. We have suggested that a useful 357
approach is to synthesize neuroanatomical, neurophysiological, and behavioral evidence, 358
using humans as a benchmark. We recognize that a distinction between primary and 359
higher-order consciousness implies that mechanisms underlying putative primary 360
conscious states might be distinct from, though possibly overlapping with, mechanisms 361
allowing its accurate report, such that absence of evidence need not be evidence of 362
absence in regard to animal consciousness. 363

Page 13

364
Within this framework, we find that birds exhibit rich cognitive and behavioral 365
capabilities consistent with conscious states, including working memory, social learning, 366
planning, and possibly even insight during problem solving. These capabilities are 367
complemented by substantial anatomical homologies and functional similarities with 368
mammals in the thalamocortical systems that are associated with consciousness. The 369
case for cephalopod molluscs is currently much less clear. However, abundant evidence 370
of sophisticated learning and memory faculties and rich behaviors, as well as early 371
indications from studies of cephalopod neurophysiology, suggest at least the possibility 372
of conscious states. 373
374
Given profound gaps that remain in the neuroanatomical characterization of both 375
subphyla, many basic questions remain concerning the existence, form, and prevalence of 376
non-mammalian consciousness (see Outstanding Questions). Future progress in 377
addressing these questions will require elaboration of behavioral paradigms designed to 378
assess complex discriminatory behavior associated with consciousness. Vocal learning in 379
birds provides a particularly promising avenue for achieving this objective (Box 1). In all 380
cases, theoretical developments are necessary to facilitate the transition from correlation
to causal explanation 14. Such a transition will allow attribution of animal consciousness
381
382
to be based causally on neural properties rather than on indirect behavioral report. 383
Finally, we note that work on animal consciousness may help in assessing consciousness 384
in humans incapable of report, including infants and patients in vegetative and minimally 385
conscious states. 386
387
Glossary 388
389
Accurate report is a first-person account of what an individual is experiencing, made 390
without the attempt to mislead. Accurate report, which can be given through language or 391
related varieties of voluntary response, has been critical in the investigation of conscious 392
states in humans. In animals without the faculty of natural language, forms of behavioral 393
report acting through other motor channels might be examined to determine the possible 394

Page 14

presence of high-order discriminations suggesting conscious states. 395
396
Binocular rivalry occurs when the two eyes are each simultaneously presented with a 397
different image. Rather than seeing both images superimposed on one another, the 398
subject sees one image first, then the other, in an alternating sequence. For example, if 399
one eye is presented with parallel vertical stripes and the other with horizontal parallel 400
stripes, rather than seeing an overlapping ‘weave’ of vertical and horizontal stripes, the
subject sees first one orientation of stripes, then the other 83.
401
402
403
Explanatory correlates of consciousness. Conventional approaches within 404
consciousness science have emphasized the search for so-called ‘neural correlates of 405
consciousness’: neural activity having privileged status in the generation of conscious
experience 7. However, the transition from correlation to explanation requires an
406
407
explanation of how particular neural correlates account for specific properties of 408
consciousness. Searches for explanatory correlates of consciousness attempt to provide
this link 11 14.
409
410
411
Primary consciousness refers to the experience of a multimodal scene composed of 412
basic perceptual and motor events. Primary consciousness is sometimes called perceptual 413
or phenomenal consciousness, and it may be present in animals without true language. 414
By contrast, higher-order consciousness involves the referral of the contents of primary 415
consciousness to interpretative semantics, including a sense of self and, in more advanced
forms, the ability to explicitly construct past and future narratives 3. The presence of
416
417
higher order consciousness or metacognition should not be assumed to be necessary for 418
the ascription of primary consciousness, though it may be constitutively required for 419
advanced forms of self-consciousness and consciousness of consciousness (Box 2). 420
421
Transitive inference is the ability to connect two or more separate relations, and is 422
widely regarded as a fundamental process in reasoning. Such abilities have been 423
demonstrated in certain birds. For example, in experiments involving the relative ranking 424
of objects presented in pairs, pigeons were able to determine that B>D after they had 425

Page 15

separately learned that A>B, B>C, C>D, and D>E, while great tits have been observed to
deduce complex social dominance rankings 30.
426
427
428