No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Studying Introspection in Animals and AIs
Abstract
The study of introspection has, up until now, been predominantly human-centric, with regrettably little attention devoted to the question of whether introspection might exist in non-humans, such as animals and artificial intelligence (AI), and what distinct forms it might take. In their target article, Kammerer and Frankish (this issue) aim to address this oversight by offering a non-anthropocentric framework for understanding introspection that could be used to address these questions. However, their discussions on introspection in animals and AIs were quite brief. In this commentary, we will build on their suggestions to offer some methodological guidance for how future research into introspection in animals and AIs might proceed.
... In our pursuit of defining the concept of introspection, our approach is closely related to the research programme proposed by Kammerer and Frankish (2023), which opens the application of the concept of introspection to nonhuman entities including artificial systems and animals. Browning and Veit (2023) refined this approach by proposing a focus on the intended design of the system to assess its introspection-like capabilities. They illustrated this by hypothesising that robots with social goals may need to introspect about their own internal states in order to understand the human beings they interact with. ...
Large language models (LLMs) exhibit compelling linguistic behaviour, and sometimes offer self-reports, that is to say statements about their own nature, inner workings, or behaviour. In humans, such reports are often attributed to a faculty of introspection and are typically linked to consciousness. This raises the question of how to interpret self-reports produced by LLMs, given their increasing linguistic fluency and cognitive capabilities. To what extent (if any) can the concept of introspection be meaningfully applied to LLMs? Here, we present and critique two examples of apparent introspective self-report from LLMs. In the first example, an LLM attempts to describe the process behind its own ``creative'' writing, and we argue this is not a valid example of introspection. In the second example, an LLM correctly infers the value of its own temperature parameter, and we argue that this can be legitimately considered a minimal example of introspection, albeit one that is (presumably) not accompanied by conscious experience.
... I also offer a summary of the core arguments of the book and assess whether we have reached the goals I set out to meet in the first chapter. Finally, the book ends with further reflections on how the pathological complexity framework could be refined, how we can use it to improve animal welfare, and even how animal minds research may help us to understand AI minds in the future (see also Browning & Veit, 2023). Much work, of course, remains to be done and I hope that the book can help us to take the first steps towards a true Darwinian study of consciousness in which animal cognition researchers and cognitive ethologists come to uncover the subjective worlds of the animals we share this planet with. ...
A Philosophy for the Science of Animal Consciousness aims to advance Donald Griffin’s vision of the “final, crowning chapter of the Darwinian revolution” by firmly integrating animals within the science of consciousness. Although this field has largely neglected the questions of when and why consciousness evolved, this book champions a Darwinian philosophy where the experiences of other animals are put centre-stage in investigations of consciousness. This synopsis offers a summary of the book’s core arguments for the advancement of a truly biological science of consciousness. This approach allows for an empirically rigorous investigation into what it is like to be a bat, a crow, a bee, or an octopus.
... This concept may require introspective capacities, and may thus surprisingly be applied to non-human animals and AIs(Browning and Veit 2023).3 Unsurprisingly, philosophers have been among those who have criticized the rational choice axiom of transitive preferences early on(Schumm 1987). ...
The axiom of transitivity has been challenged in economic theorizing for over seventy years. Yet, there does not seem to be any movement in economics towards removing classical rational choice models from introductory microeconomics books. The concept of rationality has similarly been employed in the cognitive sciences and biology, and yet, transitivity has here not only been shown to be violated, but also rationally so. Some economists have thus responded with attempts to develop alternative theories that give up on the axiom of transitivity. In this paper, I argue that there is a conceptual confusion in this debate that rests on the mistaken idea that there is something like the “one true theory of rationality” that can determine axioms like transitivity to be true or false. Instead, I defend a shift towards a pluralism of concepts of rationality as well as models in which transitivity should play a role depending on the purposes of the model at hand.
This article challenges conventional boundaries between human and artificial cognition by examining introspective capabilities in large language models (LLMs). Although humans have traditionally been considered unique in their ability to reflect on their own mental states, we argue that LLMs may not only possess genuine introspective abilities but potentially excel at them compared to humans. We discuss five objections to machine introspection: (1) the lack of direct routes to self‐knowledge in training data, (2) the conflict between static knowledge and dynamic mental states, (3) the distorting effects of reinforcement learning on self‐reports, (4) LLMs own denials of inner experience, and (5) arguments that LLMs simply mimic language without understanding. We think all these arguments fail and that there are deep parallels between human and machine introspection. Most provocatively, we propose that LLMs superior processing capabilities and pattern recognition may enable them to develop more sophisticated theories of mind than humans possess, potentially making them more reliable introspectors than their creators. If we are right, this has significant implications for artificial intelligence (AI) alignment, transparency, and our understanding of the nature of AI.
What is the role of consciousness in nature? The science of consciousness has largely neglected the question through its emphasis on human experience. In this précis of A Philosophy for the Science of Animal Consciousness, I outline how we can move from a top-down approach that begins with investigations in humans to an evolutionary bottom-up approach that targets the adaptive origins of even the most minimal forms of subjective experience. I will also offer an introduction to the central thesis of the book, that is, the pathological complexity thesis, according to which consciousness evolved in order to enable animals to adaptively respond to their life history challenges.
Measurement of subjective animal welfare creates a special problem in validating the measurement indicators used. Validation is required to ensure indicators are measuring the intended target state, and not some other object. While indicators can usually be validated through looking for correlation between target and indicator under controlled manipulations, this is not possible when the target state is not directly accessible. In this paper, I outline a four-step approach using the concept of robustness, that can help with validating indicators of subjective animal welfare.
Why is it that people simultaneously treat social robots as mere designed artifacts, yet show willingness to interact with them as if they were real agents? Here, we argue that Dennett’s distinction between the intentional stance and the design stance can help us to resolve this puzzle, allowing us to further our understanding of social robots as interactive depictions.
This article introduces and defends the “pathological complexity thesis” as a hypothesis about the evolutionary origins of minimal consciousness, or sentience, that connects the study of animal consciousness closely with work in behavioral ecology and evolutionary biology. I argue that consciousness is an adaptive solution to a design problem that led to the extinction of complex multicellular animal life following the Avalon explosion and that was subsequently solved during the Cambrian explosion. This is the economic trade-off problem of having to deal with a complex body with high degrees of freedom, what I call “pathological complexity.” By modeling the explosion of this computational complexity using the resources of state-based behavioral and life history theory we will be able to provide an evolutionary bottom-up framework to make sense of subjective experience and its function in nature by paying close attention to the ecological lifestyles of different animals.
In order to develop a true biological science of consciousness, we have to remove humans from the centre of reference and develop a bottom-up comparative study of animal minds as Donald Griffin intended with his call for a ‘cognitive ethology’. In this article, I make use of the pathological complexity thesis (Veit 2022a,c,b) to show that we can firmly ground a comparative study of animal consciousness by drawing on the resources of state-based behavioral life-history theory. By comparing the different life histories of gastropods and arthropods, we will be able to make better sense of the possible origins of consciousness and its function for organisms in their natural environments.
In the last decade, the study of magic has started to gain the attention of the psychological sciences (Kuhn, 2019; Tompkins, 2019). Psychologists study magic to further our understanding of human cognition including perception, memory, and consciousness. Here, we explore the conscious elements that might make magic unique to humans, such as the experience of wonder and disbelief at the unexpected. We discuss the tactics used by cognitively advanced nonhuman animals such as apes, corvids, and cephalopods to deceive conspecifics, and consider the parallels between their techniques and the ones used by magicians to make their audience experience the impossible. We also discuss the social dynamics of magic performance, explore the similarities and differences of human and nonhuman social interactions, and discuss play behavior in nonhuman animals. We argue that apes, corvids, and cephalopods might be ideal candidates to start a comparative science of magic, as they appear to naturally exploit analogous blind spots in perception and attention, as well roadblocks in memory, mental time travel, and perspective-taking. We highlight the potential of this new and exciting line of research that unlocks alternative avenues for inquiry and investigation. The application of magic to comparative psychology might reveal several interesting psychological constraints across diverse animal minds and offer potential candidates for questions about consciousness.
Sentience is the capacity to have feelings, such as feelings of pain, pleasure, hunger, thirst, warmth, joy, comfort and excitement. It is not simply the capacity to feel pain, but feelings of pain, distress or harm, broadly understood, have a special significance for animal welfare law.
Drawing on over 300 scientific studies, we have evaluated the evidence of sentience in two groups of invertebrate animals: the cephalopod molluscs or, for short, cephalopods (including octopods, squid and cuttlefish) and the decapod crustaceans or, for short, decapods (including crabs, lobsters and crayfish). We have also evaluated the potential welfare implications of current commercial practices involving these animals.
How does consciousness vary across the animal kingdom? Are some animals ‘more conscious’ than others? This article presents a multidimensional framework for understanding interspecies variation in states of consciousness. The framework distinguishes five key dimensions of variation: perceptual richness, evaluative richness, integration at a time, integration across time, and self-consciousness. For each dimension, existing experiments that bear on it are reviewed and future experiments are suggested. By assessing a given species against each dimension, we can construct a consciousness profile for that species. On this framework, there is no single scale along which species can be ranked as more or less conscious. Rather, each species has its own distinctive consciousness profile.
In 2009, a special issue of Comparative Cognition and Behavior Reviews offered a series of target articles by researchers and theorists who debated the claim that animals were capable of engaging in metacognition. That special issue outlined some of the crucial points of contention at that time, and it also pointed the way toward refinements in the approaches used in empirical work with animals. The papers in that special issue framed the debate about whether animals’ performances were better accounted for by learning mechanisms or first-order cognitive processes rather than metacognitive processes. In the decade since that special issue, substantial empirical and theoretical work has emerged on this question. This issue of Animal Behavior and Cognition presents some of that ongoing research and the new debate that it has generated. Proponents for and against claims that animals engage in metacognition offer their perspectives on existing data and what future research is needed to continue to move forward the exploration of animal metacognition. Research of the past decade has demonstrated a range of behavioral phenomena that cannot be accounted for by simply appealing to associative learning or monitoring external stimuli that serve as cues to evoke escape responses and other forms of metacognitive behaviors. However, there remain some important empirical and theoretical challenges to accepting that some animals are metacognitive (as was true also in 2009). These challenges should help frame the next decade of research, while at the same time being responsive to the full range of reported competencies shown by animals to date, and into the future.
Fiery Cushman argues that “[r]ationalization is designed not to accurately infer unconscious mental states, but to construct new ones; it is not a discovery, but a fiction”. While we agree in broad strokes with the characterization of rationalization as a ‘useful fiction’, we think that Cushman’s claim remains ambiguous in two crucial respects: (i) the reality of beliefs and desires, i.e. the fictional status of folk psychological entities, and (ii) the degree to which they should be understood as useful and representative. Our aim here is to clarify both points and illuminate how rationalization could be understood as a useful fiction. In doing so, we aim to explicate the Rationale of Rationalization.
Knowing the extent to which nonhumans and humans share mechanisms for metacognition will advance our understanding of cognitive evolution and will improve selection of model systems for biomedical research. Some nonhuman species avoid difficult cognitive tests, seek information when ignorant, or otherwise behave in ways consistent with metacognition. There is agreement that some nonhuman animals "succeed" in these metacognitive tasks, but little consensus about the cognitive mechanisms underlying performance. In one paradigm, rhesus monkeys visually searched for hidden food when ignorant of the location of the food, but acted immediately when knowledgeable. This result has been interpreted as evidence that monkeys introspectively monitored their memory to adaptively control information seeking. However, convincing alternative hypotheses have been advanced that might also account for the adaptive pattern of visual searching. We evaluated seven hypotheses using a computerized task in which monkeys chose either to take memory tests immediately or to see the answer again before proceeding to the test. We found no evidence to support the hypotheses of behavioral cue association, rote response learning, expectancy violation, response competition, generalized search strategy, or postural mediation. In contrast, we repeatedly found evidence to support the memory monitoring hypothesis. Monkeys chose to see the answer when memory was poor, either from natural variation or experimental manipulation. We found limited evidence that monkeys also monitored the fluency of memory access. Overall, the evidence indicates that rhesus monkeys can use memory strength as a discriminative cue for information seeking, consistent with introspective monitoring of explicit memory. (PsycINFO Database Record (c) 2014 APA, all rights reserved).
Animals have to process quantity of information in order to take decisions and adapt their behaviors to their physical and social environment. They have to remember previous events (learning), to cope with their internal (motivational and emotional) states and to display flexible behavioral responses. From a human point of view it is quite impossible to access all those information, not only because of the sensorial channels used that can vary but also because all the pro-cessing phase occurs in the "black box" and non-human animals are not able to express verbally what they think, feel or want. Though useful information might lie in the "collected data" (animal mind), extracting them into insightful knowl-edge with human-accessible form (clear meaning, no interpretation) presents a demanding and sophisticated undertaking. Several scientists decided to trained different individuals from several species (apes, dolphins, grey parrots, dogs) in order to teach them a new communicative system that they could share with us. Here, the different studies (techniques and species used) are presented, the constrains but also the main findings.
More than 19,000 multisign utterances of an infant chimpanzee (Nim) were analyzed for syntactic and semantic regularities.
Lexical regularities were observed in the case of two-sign combinations: particular signs (for example, more) tended to occur
in a particular position. These regularities could not be attributed to memorization or to position habits, suggesting that
they were structurally constrained. That conclusion, however, was invalidated by videotape analyses, which showed that most
of Nim's utterances were prompted by his teacher's prior utterance, and that Nim interrupted his teachers to a much larger
extent than a child interrupts an adult's speech. Signed utterances of other apes (as shown on films) revealed similar non-human
patterns of discourse.
The anxiogenic nature of the interoceptive discriminative stimulus induced by pentylenetetrazole (PTZ) was investigated by examining the discriminatory behaviour of PTZ conditioned pigs during a conditioned emotional response (CER). A CER was induced in a nonoperant situation, by pairing a tone stimulus with the application of a mild, non-injurious electric shock. Subsequent presentation of the conditioned tone stimulus alone produced a generalisation to the PTZ cue. This generalisation of the conditioned emotional state (CES) to the PTZ cue was antagonised by pretreatment with diazepam (0.5 mg/kg, PO; 30 min). The PTZ stimulus was also antagonised by diazepam (0.5 mg/kg, PO; 30 min) but not by an anticonvulsant dose of ethosuximide (30 mg/kg, PO; 1-3 h), providing further confirmation of the anxiogenic nature of the PTZ cue. Our results demonstrate the validity of the PTZ discrimination paradigm in pigs as a test of anxiety.
Although mental well-being has long been accepted as an important aspect of animal welfare, the subjective feelings of farm or laboratory animals are regarded as lying beyond the scope of scientific enquiry. We now report that pharmacological conditioning of pigs with a drug, pentylenetetrazole, known to induce anxiety in man, permits investigation of the presence or absence of this psychological state during exposure to a variety of environmental stimuli encountered during normal husbandry. Such pharmacological conditioning therefore provides a valuable means to assess and improve elements of animal welfare and should be applicable to other species that show operant behaviour.
Metacognition is knowledge that can be expressed as confidence judgments about what one knows (monitoring) and by strategies for learning what one does not know (control). Although there is a substantial literature on cognitive processes in animals, little is known about their metacognitive abilities. Here we show that rhesus macaques, trained previously to make retrospective confidence judgments about their performance on perceptual tasks, transferred that ability immediately to a new perceptual task and to a working memory task. We also show that monkeys can learn to request "hints" when they are given problems that they would otherwise have to solve by trial and error. This study demonstrates, for the first time, that nonhuman primates share with humans the ability to monitor and transfer their metacognitive ability both within and between different cognitive tasks, and to seek new knowledge on a need-to-know basis.
Deep neural networks (DNNs) have found many useful applications in recent years. Of particular interest have been those instances where their successes imitate human cognition and many consider artificial intelligences to offer a lens for understanding human intelligence. Here, we criticize the underlying conflation between the predictive and explanatory power of DNNs by examining the goals of modeling.
We propose a new approach to the study of introspection. Instead of asking what form introspection actually takes in humans or other animals, we ask what forms it could take, in natural or artificial minds. What are the dimensions along which forms of introspection could vary? This is a relatively unexplored question, but it is one that has the potential to open new avenues of study and reveal new connections between existing ones. It may, for example, focus attention on possible forms of introspection radically different from the human one and help to integrate competing theories of human introspection in a non-adversarial manner. We introduce and motivate the project, provide a preliminary mapping of the space of possible forms of introspection, and sketch a programme for interdisciplinary research on the topic.
We outline a framework for evaluating scientific evidence of sentience, focusing on pain experience. It includes eight neural and cognitive-behavioural criteria, with confidence levels for each criterion reflecting the reliability and quality of the evidence. We outline the rationale for each criterion and apply our framework to a controversial sentience candidate: decapod crustaceans. We have either high or very high confidence that true crabs (infraorder Brachyura) satisfy five criteria, amounting to strong evidence of sentience. Moreover, we have high confidence that both anomuran crabs (infraorder Anomura) and astacid lobsters/crayfish (infraorder Astacidea) meet three criteria—substantial evidence of sentience. The case is, as yet, weaker for other infraorders, such as penaeid shrimps, highlighting important research gaps. Having demonstrated our framework’s application to decapod crustaceans, we hope that future research will apply it to other taxa.
Comparative psychology, the multidisciplinary study of animal behavior and psychology, confronts the challenge of how to study animals we find cute and easy to anthropomorphize, and animals we find odd and easy to objectify, without letting these biases negatively impact the science. In this Element, Kristin Andrews identifies and critically examines the principles of comparative psychology and shows how they can introduce other biases by objectifying animal subjects and encouraging scientists to remain detached. Andrews outlines the scientific benefits of treating animals as sentient research participants who come from their own social contexts and with whom we will be in relationship. With discussions of science's quest for objectivity, worries about romantic and killjoy theories, and debates about chimpanzee cognition between primatologists who work in the field and those in the lab, Andrews shows how scientists can address the different biases through greater integration of the subdisciplines of comparative psychology.
Metacognition is the ability to monitor and control one’s cognition. Monitoring may involve either public cues or introspection of private cognitive states. We tested rhesus monkeys (Macaca mulatta) in a series of generalization tests to determine which type of cues control metacognition. In Experiment 1, monkeys learned a perceptual discrimination in which a “decline-test” response allowed them to avoid tests and receive a guaranteed small reward. Monkeys declined more difficult than easy tests. In Experiments 2-4, we evaluated whether monkeys generalized this metacognitive responding to new perceptual tests. Monkeys showed a trend toward generalization in Experiments 2 & 3, and reliable generalization in Experiment 4. In Experiments 5 & 6, we presented the decline-test response in a delayed matching-to-sample task. Memory tests differed from perceptual tests in that the appearance of the test display could not control metacognitive responding. In Experiment 6, monkeys made prospective metamemory judgments before seeing the tests. Generalization across perceptual tests with different visual properties and mixed generalization from perceptual to memory tests provide provisional evidence that domain-general, private cues controlled metacognition in some monkeys. We observed individual differences in generalization, suggesting that monkeys differ in use of public and private metacognitive cues.
The extent to which nonhuman animals can learn actual human language is a controversial question, but many nonhuman species have acquired elements of a two-way communication system that is, and was, sophisticated enough to enable its use in evaluating cognitive capacities. This article is a personal view of the history of these animal language studies.
The chimpanzee as a subject. Some discussion of the chimpanzee as an experimental subject is in order because this species is relatively uncommon in the psychological laboratory. Whether or not the chimpanzee is the most in- telligent animal after man can be dis- puted; the gorilla, the orangutan, and even the dolphin have their loyal par- tisans in this debate. Nevertheless, it is generally conceded that chimpanzees are highly intelligent, and that mem- bers of this species might be intelligent enough for our purposes. Of equal or greater importance is their sociability and their capacity for forming strong attachments to human beings. We want to emphasize this trait of sociability; it seems highly likely that it is essential for the development of language in human beings, and it was a primary consideration in our choice of a chim- panzee as a subject.
The intentional stance is the strategy of prediction and explanation that attributes beliefs, desires, and other “intentional” states to systems – living and nonliving – and predicts future behavior from what it would be rational for an agent to do, given those beliefs and desires. Any system whose performance can be thus predicted and explained is an intentional system, whatever its innards. The strategy of treating parts of the world as intentional systems is the foundation of “folk psychology,” but is also exploited (and is virtually unavoidable) in artificial intelligence and cognitive science more generally, as well as in evolutionary theory. An analysis of the role of the intentional stance and its presuppositions supports a naturalistic theory of mental states and events, their content or intentionality, and the relation between “mentalistic” levels of explanation and neurophysiological or mechanistic levels of explanation. As such, the analysis of the intentional stance grounds a theory of the mind and its relation to the body.
The possibility that memory awareness occurs in nonhuman animals has been evaluated by providing opportunity to decline memory tests. Current evidence suggests that rhesus monkeys (Macaca mulatta) selectively decline tests when memory is weak (Hampton in Proc Natl Acad Sci USA 98:5359-5362, 2001; Smith et al. in Behav Brain Sci 26:317-374, 2003). However, much of the existing research in nonhuman metacognition is subject to the criticism that, after considerable training on one test type, subjects learn to decline difficult trials based on associative learning of external test-specific contingencies rather than by evaluating the private status of memory or other cognitive states. We evaluated whether such test-specific associations could account for performance by presenting monkeys with a series of generalization tests across which no single association with external stimuli was likely to adaptively control use of the decline response. Six monkeys performed a four alternative delayed matching to location task and were significantly more accurate on trials with a decline option available than on trials without it, indicating that subjects selectively declined tests when memory was weak. Monkeys transferred appropriate use of the decline response under three conditions that assessed generalization: two tests that weakened memory and one test that enhanced memory in a novel way. Bidirectional generalization indicates that use of the decline response by monkeys is not controlled by specific external stimuli but is rather a flexible behavior based on a private assessment of memory.
A female lowland gorilla, Koko, has been engaged in an ongoing language program since July 1972 when she was 1 year old. During the first 30 months of training she acquired a vocabulary of 100 words in American Sign Language which she spontaneously combined into meaningful and often novel statements of up to 11 signs in length. The gorilla is using a rapidly expanding vocabulary of signs to express semantic and possibly grammatical relations similar to those expressed by human children in the early stages of language acquisition. Patterns of generalization, gradual increase in mean length of utterance, and innovative use of gestural language are discussed in relation to data available on children and chimpanzees.
Using a novel pharmacological conditioning procedure, pigs were conditioned to discriminate the effects of a subconvulsant dose of the anxiogenic drug pentylenetetrazole (PTZ; 2.8-10 mg/kg, i.v.) from saline. The operant chamber provided two levers at which pigs were trained to press at a fixed ratio of 20 presses per food reward (FR 20). The animals were conditioned to select both levers alternately following saline treatment and to select one lever only following PTZ treatment. This procedure enabled the onset and offset of the PTZ stimulus to be detected within single test sessions; infusion of PTZ to animals already selecting both levers alternately after a saline pretreatment induced a period of response exclusive to the PTZ lever followed by a return to an alternation of lever selection response. The ability of the novel procedure to detect the time course of the drug cue should improve future evaluations of the psychological states induced by centrally acting drugs. With PTZ as the training drug, the novel procedure presents a valuable means to study the neurobiology of anxiety.
As a consequence of domestication, dogs have a special readiness for communication with humans. We here investigate whether a dog might be able to acquire and consistently produce a set of arbitrary signs in her communication with humans, as was demonstrated in "linguistic" individuals of several species. A female mongrel dog was submitted to a training schedule in which, after basic command training and after acquiring the verbal labels of rewarding objects or activities, she learned to ask for such objects or activities by selecting lexigrams and pressing keys on a keyboard. Systematic records taken during spontaneous interaction with one of the experimenters showed that lexigrams were used in an appropriate, intentional way, in accordance with the immediate motivational context. The dog only utilized the keyboard in the experimenter's presence and gazed to him more frequently after key pressing than before, an indication that lexigram use did have communicative content. Results suggest that dogs may be able to learn a conventional system of signs associated to specific objects and activities, functionally analogous to spontaneous soliciting behaviors and point to the potential fruitfulness of the keyboard/lexigram procedure for studying dog communication and cognition. This is the first report to systematically analyze the learning of arbitrary sign production in dogs.
Next of Kin: My Conversations with Chimpanzees
- R Fouts
- S Mills
Fouts, R. & Mills, S. (1997) Next of Kin: My Conversations with Chimpanzees,
New York: Avon Books.
Can insects feel pain? A review of the neural and behavioural evidence
- M Gibbons
- A Crump
- M Barrett
- S Sarlak
- J Birch
- L Chittka
Gibbons, M., Crump, A., Barrett, M., Sarlak, S., Birch, J. & Chittka, L. (2022) Can
insects feel pain? A review of the neural and behavioural evidence, in Jurenka,
R. (ed.) Advances in Insect Physiology, vol. 63, pp. 155-229, Amsterdam:
Elsevier.
How Stella Learned to Talk: The Groundbreaking Story of the World's First Talking Dog
- C Hunger
Hunger, C. (2021) How Stella Learned to Talk: The Groundbreaking Story of the
World's First Talking Dog, New York: William Morrow.
Language' and Intelligence in Monkeys and Apes: Comparative Developmental Perspectives
- H L W Miles
Miles, H.L.W. (1990) The cognitive foundations for reference in a signing orangetan, in Taylor Parker, S. & Gibson, K.R. (eds.) 'Language' and Intelligence in
Monkeys and Apes: Comparative Developmental Perspectives, pp. 511-539,
Cambridge: Cambridge University Press.
Language Learning by a Chimpanzee: The Lana Project
- D M Rumbaugh
Rumbaugh, D.M. (1977) Language Learning by a Chimpanzee: The Lana Project,
New York: Academic Press.