The Emerging Science of Animal Consciousness

"What provides evidence that an animal has a single, unified perspective as opposed to multiple perspectives? It is crucial to investigate cognition as well as neuroanatomy. Here we can draw inspiration from experiments on split-brain humans. One paradigm involves training an animal to perform a task in response to a stimulus presented to one eye and seeing whether the task can still be performed when the stimulus is presented to the other eye: interocular transfer. In pigeons (Columbia livia), the visual field for each eye can be divided into two regions: the red field, which is the lower frontal region important for guiding pecking, and the yellow field, which covers the upper frontal and lateral regions. There can be interocular transfer between the red fields of each eye, but there seems to be no interocular transfer between the yellow fields in nearly all individuals [44]. Some particular individuals can do it, but no one knows why [44]."

Integration at a Time (Unity), p. 5

The interocular transfer paradigm operationalizes cross-hemispheric accessibility of learned information, aligning with information integration (unity) as a consciousness-relevant marker in biology and suggesting analogous integration tests in AI systems.

Tables

Table 1 (p. 22) : Current Experimental Paradigms for Investigating Dimensions of Animal Consciousness.

"sensory input to produce a coherent, continuous stream from discontinuous stimuli. In humans, evidence for such mechanisms comes from the colour-phi illusion, in which two spatially separated, differently coloured dots flashed in sequence are perceived as a single moving dot that changes colour halfway across the gap [49]. The brain is not simply mistaking two static stimuli for a moving stimulus: it is constructing a coherent account of how the stimulus is changing. Colour-phi has received a great deal of discussion in the philosophy of consciousness [50]. What matters here is simply that, if we found colour-phi in non-human animals, this would be evidence that they too have mechanisms that transform a series of discrete stimuli into a coherent experience of change. Although our evidence of colour-phi in humans comes from verbal report, it is possible in principle to study colour-phi in the absence of verbal criteria [51]. Animals could be trained to respond differently to perceptions of continuous and discrete stimuli and to stimuli that change colour half-way and stimuli that do not. We could then present them with a colour-phi test stimulus, gradually reducing the interstimulus interval. Would there be a threshold at which the animal switched from categorising the stimulus as discrete to categorising it as continuous and would the animal categorise the stimulus as one that changes colour half-way?"

Integration across Time (Temporality), p. 6

This passage links temporal binding mechanisms to conscious stream formation and outlines a nonverbal paradigm for animals, paralleling temporal coordination concepts that could be probed in AI via timing and synchronization analyses.

Figures

Figure 1 (p. 20) : Highlights temporality as a distinct dimension, motivating temporal-coordination tests across species and potentially in AI benchmarks.

"Experiments probing motivational trade-offs can provide insight into how evaluation systems vary. In one such experiment, rats (Rattus norvegicus domestica) were presented with an opportunity to access a sugar solution by entering a cold chamber [30]. The rats traded off the sugar content of the solution against the temperature of the chamber: all else being equal, they were willing to withstand colder temperatures to get sweeter rewards. This is evidence of an evaluative common currency: the value of sugar is weighed subtly against the disvalue of cold. Is it also evidence of conscious experience of the currency? It is relevant that the trade-off is crossmodal: this is not an animal evaluating options using information from a single sense, but an animal weighing the taste of a liquid against the temperature of the ambient environment. This requires the crossmodal integration of information, which has often been linked to consciousness, although it may not strictly require it [31,32]."

E-Richness, p. 4

Crossmodal cost–benefit trade-offs provide behavioral access to valence processing potentially relevant to welfare and conscious affect, suggesting analogous negative-value channels or persistence measures to probe in AI agents.

Tables

Table 1 (p. 22) : Current Experimental Paradigms for Investigating Dimensions of Animal Consciousness.

"A more sophisticated grade of self-consciousness involves awareness of one’s own body as a persisting object that exists in the world [65]. This capacity is plausibly needed to pass a mirror-mark test, in which the test subject is able to recognise a mark seen in a mirror as a mark on its own body. Chimpanzees (Pan troglodytes) [66], bottlenose dolphins (Tursiops truncatus) [67], Asian elephants (Elephas maximus) [68], and magpies (Pica pica) [69] have reportedly passed such a test. A striking study in 2019 reported that a fish, the cleaner wrasse (Labroides dimidiatus), can also pass the test [70]. Fish able to view a coloured mark on their throat in the mirror were much more likely to exhibit throat-scraping behaviour, as if to remove a parasite, than fish who had transparent marks or no access to a mirror. These results are controversial [71], but they suggest that the grade of self-consciousness required to pass the mirror-mark test is possessed by a wide range of animals."

Self-Consciousness (Selfhood), p. 7

Mirror-mark performance operationalizes a report-like pathway for self-referential content, informing how self-model and reportability might be dissociated or aligned in AI systems with explicit introspection modules.