Are animals self-aware?

A review of “The missing link in cognition: Origins of self-reflective consciousness”

Herbert Terrace and Janet Metcalfe assemble thirteen essays on how self-awareness evolved and up to what point nonhuman organisms possess this ability. Their book is highly informative and interesting, but the missing link is still… missing.

Introduction

“The central issue in this volume is where self-reflective consciousness should be situated among the numberless gradations hypothesized by Darwin. One can ask, perhaps too simplistically, did self-reflective consciousness evolve before or after language? A better question would be, which gradations of self-reflective consciousness require language and which do not?” (p. xi).

These are fascinating questions indeed, but despite honest efforts and innovative experiments presented by contributors to Terrace and Metcalfe’s book, they mostly remain unanswered. One main unavoidable obstacle is that pure, genuine self-reflective consciousness cannot be directly tested in non-verbal organisms, so that lesser manifestations of it need to be used instead. We end up with a rich set of evidence indicating that birds, dolphins, primates, and babies do exhibit some of these watered-down versions of self-reflection, but this tells us little about real “self-reflective consciousness”—“the ability to doubt what one knows, to deny or affirm one’s beliefs, to judge one’s own memories and percepts, to comment on one’s dreams, to recollect and reflect upon one’s own past…”, as defined by the editors in their opening chapter (p. xii)—curiously unlisted in the Table of Contents.

What are these abilities used to infer the presence of self-awareness in non-verbal creatures? Contributors to this volume mainly focus on autonoetic consciousness (the capacity to remember one’s own past and to imagine one’s possible future) and metacognition (the ability to monitor one’s mental states). There is a strong emphasis on evolutionary theory and primate cognition throughout the book. Four chapters are devoted to autonoetic consciousness and three examine metacognition; the remaining six chapters either discuss these together with other operationalizations of self-reflective consciousness (e.g., gaze following, self-recognition, Theory of Mind) or look at other aspects of the problem (e.g., levels of consciousness, the importance of language, evolutionary functions of self-awareness).

I start with some general observations about the book and then turn my attention to specific chapters.

Overview

The book (published in 2005) offers a collection of contributions originally presented at a Columbia University conference held in April 2002. I was fearful at first that the content would thus be somewhat dated, but all authors updated their original talks with recent (2004) references. The editors have organized the chapters in a logical fashion respecting the epigenetic principle, where each contribution builds on the previous one and sets the stage for the next one—in most cases, anyway. It may be tempting to skip some chapters and directly jump to those one finds more interesting, but I suggest it is better to read the book from cover to cover, following its intended structure. Almost all contributors refer to work presented in other chapters, which gives coherence and continuity to the book.

This is not an easy book to read and the target audience most certainly consists of researchers with a solid background in the areas covered. However, it is not overly specialized to the point of being opaque, perhaps with the exceptions of chapters 3 (Herbert Terrace) and 7 (Patricia Kitcher), that I found particularly difficult to grasp. (This obviously might have been the result of my own intellectual limitations.) In this first sense the book may not be that well One main unavoidable obstacle is that pure, genuine self-reflective consciousness cannot be directly tested in non-verbal organisms. “balanced”, meaning that contributions vary in complexity. Note that papers on primate cognition (chapters 8, 9, 10, 11, and 12) tend to be fairly technical. I believe that the book is not very well balanced in a second sense as well—the “quality” (i.e., depth, originality) of chapters is fairly unequal, with some outstanding reports (e.g., chapters 1, 6, 9, and 13, by Endel Tulving, E. Tory Higgins, Bennett L. Schwartz, and Joseph Call, respectively) and weaker contributions (chapters 2 and 7 by Janet Metcalfe/Hedy Kober and Patricia Kitcher unfortunately come to mind). My more specific comments to be made below will explicate these ratings. In yet another sense however, the book indeed seems well balanced with about half of all contributors suggesting that self-reflective consciousness can only be found in humans, and the other half claiming that nonhuman organisms possess at least some precursors of self-awareness.

One final observation before we look at some of the evidence presented in the book. This might sound somewhat trivial, but I question the usefulness of having included a set of color illustrations in the middle of the book. These images, pictures, charts, and graphs are already displayed (in black and white) in the appropriate chapters, and most probably did inflate the purchase price ($67.50 USD on Amazon, January 2006).

Autonoetic Consciousness

Autonoetic consciousness is the focus of the first and longest chapter in the book, written by Endel Tulving in a particularly clear, careful, and engaging style. “It ought to be completely noncontroversial to say that nonhuman animals are conscious, that they have memory, that they can learn, that they can think, that they know what happens in their world—in brief, that they are ‘intelligent’…”. But, Tulving then adds, “… Human beings possess a form of It ought to be completely noncontroversial to say that nonhuman animals are conscious. memory (episodic memory) and a form of consciousness (autonoetic consciousness, or ‘autonoesis’) that no other animals do… They do not exist in insects, in birds, in mice or rats, in cats or dogs, and not even in gorillas and chimps” (p. 5).Autonoetic consciousness is mental time travel that “… takes the form of remembering personally experienced and thought-about events, occasions, and situations that occured in the past, together with imagining (preexperiencing) personal happenings in the subjectively felt future” (p. 47). By definition autonoetic consciousness is part of the broader self-reflective consciousness because it involves one reflecting on one’s past experiences and possible future. It is the felt experience that it is I who went through specific past events and will encounter similar and possibly different situations in the future. The ability to imagine our future represents a crucial part of autonoetic consciousness. Needless to say, “… a staggering large portion of human behavior today—social, economic, political, religious, and otherwise—is governed… by an awareness of the future” (p. 40).

Autonoetic consciousness is directly linked to (but is more than) episodic memory, the ability to remember episodes of one’s own life, or the capacity to recollect individual events from one’s personal life. Episodic memory has to be distinguished from semantic memory, the capacity to know things about the world. So if I ask you to tell me who the president of the United States is, you should readily answer “George Bush”—this is semantic memory, information about the world stored in your memory. If I ask you the name of your parents, or the location of your house, or the make of car you currently drive, there too you should be able to respond without hesitation. Does recall of food locations and memory of spatial relationships mean self-reflective consciousness? Note that although these last examples pertain to yourself, they still illustrate semantic memory, because the recall of specific personal episodes is not yet necessary. In other words, you should easily be able to recollect the name of your parents without having to consciously remember individual past interactions you had with them. If, however, I indeed ask you to do so, or to recount your first date with your current partner, you should be able to produce a narrative with specific details describing that episode in your life. This is episodic memory—being capable of mentally reliving the past.

In his chapter, Tulving closely examines the relevant evidence regarding episodic memory and autonoetic consciousness in animals, and concludes (as stated earlier) that they do not exhibit these abilities. This also applies to young children, who can recollect facts about their own lives (e.g., remembering what they put in a drawer a few weeks ago—semantic memory) but not personal events (the actually memory of them putting stuff in the drawer—episodic memory). Autonoetic consciousness would gradually develop around the age of four and would be greatly facilitated by the acquisition of language.

Both Charles Menzel (chapter 8) and Bennett L. Schwartz (chapter 9) disagree with Tulving about the absence of episodic memory in animals and present results of experiments to support their claim. Menzel reports his own observations on primates’ recall abilities. A typical study goes as follows (I simplify his research design and procedure here, as these are actually very technical and complex): a chimp is first allowed to witness the experimenter hiding food at various locations in an outdoor enclosure; the following day the chimp is invited to go food hunting in the enclosure, accompanied by an uninformed caregiver whose task is to recover the hidden food by responding to the chimp pointing toward various locations. That kind of delayed-response procedure in a loosely structured and natural environment shows that primates are remarkably quick and accurate in recalling where the rewards have been hidden; they also tend to take the shortest route from one food location to the next (as opposed to aimlessly wandering in the enclosure), which suggests a memory of spatial relationships among locations as well

Nice. But one is tempted to say—so what? Does recall of food locations and memory of spatial relationships mean self-reflective consciousness? Schwartz in the next chapter offers this thought: “Menzel’s study represents a significant advance in the study of nonhuman episodic memory, but there is a problem when interpreting Menzel’s results as evidence of episodic memory. The chimpanzee’s memory need not have been [episodic]. It is possible that [the subject] updated her memory about spatial landmarks and their contents and did not need to mentally refer to the event of watching [the experimenter] hiding the food. In this view, retrieval would be based on the present state of the world and not on the occurence of a past event” (p. 231).

Schwartz also describes his own experiments with the gorilla King. He first indicates that the distinct feeling of pastness so central to episodic memory can easily be assessed in people by verbally asking “What do you recall?” or “What happened to you?”, and acknowledges that “Obviously, these kinds of verbal tests cannot be conducted with animals…” (p. 228), so a non-verbal operational definition of episodic memory needs to be formulated. “The ape’s response must be based on retrieval from long-term memory of a unique event that was learned in a single trial. In A non-verbal operational definition of episodic memory needs to be formulated. addition, the animal’s response should provide information about its past rather than about the current state of its knowledge” (p. 232). I present one simple example from a series of studies. King would first witness a unique event—e.g., seeing a familiar person doing something odd, such as stealing a cell phone. This was followed by a 15-minute retention interval, and then King was shown three photographs (two distractors and the witnessed event) and was asked to select the correct photograph depicting what he had seen before. A good answer was rewarded with food. Although not always accurate, King performed significantly higher than chance on that type of trial. Schwartz concludes: “Because King’s memory attains the criterion of… reference to the past, I think we may claim to have supported the contention that King has the rudiments of an episodic memory system” (p. 237). The key word here is rudiments. This is a good example of a researcher having to reduce an aspect of self-reflective consciousness to its simplest expression in order to test its presence or absence in non-verbal animals. Schwartz certainly can’t be blamed for that—but we nonetheless end up with a weak indicator of episodic memory (and self-awareness).

Metacognition

The second main cognitive ability used by some contributors to study self-reflective consciousness in non-verbal organisms is metacognition. Metacognition refers to “… thinking about thinking, or cognition about cognition. The idea in this field is that in some minds mental activities occur at a higher ‘metalevel’ and at a lower ‘object level’ during cognitive processing. In these minds there is a cognitive executive that supervises (i.e., oversees and facilitates) thought or problem solving” (p. 242). Metacognition represents a specific form of self-reflective consciousness; whereas the later entails the ability to reflect on any aspect of the self (e.g., personality traits, values, opinions, emotions, behavior, physical appearance), metacognition specializes in awareness of one’s mental operations. In humans, this means becoming aware of a thought one just had (“I just thought about all these exams I still have to mark”) or how one just solved a problem (“I simply needed to carefully read the instructions manual”). One other instance of metacognition is when we feel uncertain about some information we might possess—or not. When you think “Can I recall this phone number or do I need to look it up in the directory?”, you are actively probing your memory system and thus are reflecting on your cognition. When you ask yourself “Did I hear the phone ringing or was it just my imagination?”, you also are reflecting on your thought processes (in this case, a potential perceptual experience).

Many contributors in Terrace and Metcalfe’s volume use the “uncertainty response” to assess metacognitive abilities in nonhuman animals. In humans the uncertainty response can be observed in psychophysical tasks, when participants are asked to discriminate between two tones, for example. When the task becomes too difficult (i.e., when the two tones become almost identical), people will Because an animal demonstrates self-reflective cognition in one domain does not necessarily mean that it possesses it in other domains. spontaneously report being uncertain. The experimenter can objectively measure the uncertainty response by offering three possible options to participants: (1) the second tone is higher than the first one, (2) the second tone is lower than the first one, or (3) I am unsure/uncertain. Choosing the third uncertain option can be taken as evidence of metacognitive ability, where participants are thinking about their cognitive operations (i.e., “Are the two tones different? I just don’t know”). Still in humans, the uncertainty response can also be elicited in memory tasks. For instance, participants are asked to recall a previously presented stimulus word in a list; they can respond (1) Yes it is there, (2) No it is not there, or (3) I’m uncertain, I can’t answer. This last response of declining some trials involves metacognition because participants are reflecting on their own mnemonic processes.

Similar psychophysical and memory tasks have been cleverly adapted for non-verbal animals, some of which are reported by J. David Smith (in an often humorous style) in chapter 10. Smith convincingly shows that dolphins and monkeys, but not rats, make uncertainty responses in a variety of tasks. “Given this fact, it is possible that the uncertainty response plays a metarole in their performance, too. It is possible that it is for them, too, a comment on or a response to indeterminacy and difficulty in the primary discrimination” (p. 247). Robert R. Hampton in chapter 11 presents more ape experiments on “memory awareness”. The premise is the same: “Memory awareness can make it possible to determine whether or not the information required to successfully perform a particular task is available” (p. 275). Thus, if an animal chooses to decline a memory trial because it thinks it might not remember accurately if an image has been presented or not before, it is engaged in metacognition. Hampton reports studies showing that monkeys do turn down trials because they are uncertain. Importantly, he notes that “In striking contrast to these observations of primates, pigeons tested under similar conditions appear to lack memory awareness, suggesting that it may not be widely distributed among species…” (p. 291). The author also prudently states that “Because an animal demonstrates self-reflective cognition in one domain does not necessarily mean that it possesses it in other domains” (p. 292).A variation on the “uncertainty response” theme is proposed by Lisa K. Son and Nate Kornell in chapter 12. People and animals can be asked to make “meta-confidence judgments”—i.e., to evaluate up to what point they are confident that a previously made response is correct. This requires metacognition because one has to think about one’s own knowledge (or lack thereof) when assessing one’s confidence in an answer. Knowledge, of course, is cognitive in nature. For example, I could ask you “Who painted the Sistine Chapel?”; you could answer “Michelangelo”, and then I would invite you to rate your level of confidence of being right. Son and Kornell modified this task for non-verbal animals (which I report here in an extremely simplified version) and asked monkeys to identify the longest of nine lines. The “meta-confidence judgment” part of the task consisted in subjects making a high bet (for high reinforcement) or a low bet (for lower reinforcement) on the correctness of their previous answer. Making a high bet meant that the animal was very confident in the previously given answer, whereas making a low bet indicated that it was rather unsure about the accuracy of the answer. Monkeys were indeed able to make accurate confidence judgments. The authors conclude that “… our monkeys are able to reflect on their cognitions, indicating that they possess at least one level of metacognition” (p. 318).

In the concluding chapter of the book, Joseph Call not only summarizes some key findings reported in earlier chapters but introduces his own research on metacognition. His main claim is that, besides declining a task because of lack of knowledge, organisms can also seek information when it is incomplete. If I invite you to solve a given problem and you keep asking me questions to gain more knowledge about the problem, it signifies that you know that you do no possess all the necessary information to accomplish the task. Knowing that one doesn’t know is metacognition. Call developed elegant and yet simple experimental setups to investigate information seeking (and metacognition) in human and nonhuman animals. In a typical experiment (again simplified here), subjects have to choose one of two containers to obtain a reward. The experimenter places food in one of the two containers; in one condition, subjects can see in which container the food is put before choosing; in another condition they don’t have direct visual access to that information, but if they bend down and look under the containers, they can see where the food is. Call recorded which container subjects selected, but more importantly, whether they looked under the containers before choosing—the operationalization of information seeking. Chimpanzees, gorillas, orangutans, and children actively seek additional information in that type of experiments; dogs however, do not. This contributor carefully considers alternative, non-mentalistic, explanations for his results, and cautiously concludes that metacognition seems to be involved, but “… additional research is needed to pinpoint the exact mechanism responsible for the behavior here described and its scope” (p. 336).

Other Contributions

Although autonoetic consciousness and metacognition constitute the backbone of Terrace and Metcalfe’s volume, other aspects of the problem are examined. In chapter 4, Katherine Nelson takes a human developmental perspective and rightly emphasizes that self-reflective consciousness comes in levels. (Interested readers are invited to look up my SCR paper on levels of consciousness.) She proposes six levels: physical, social, cognitive, representative/reflective, narrative, and cultural. Based on a careful review of the available evidence, her main claims are: (1) higher levels require language, and (2) “Because these language-based levels are unique to humans, we should look to the earlier levels—perhaps to cognitive consciousness—for the closest relation to our own experience among nonhuman animals…” (p. 136). About the first claim, Nelson proposes that “… conversational interactions that involve the child in reflection on self, the past, feelings, and thoughts provide the context for the representation and reflective use of language by the child, either through external private speech, or internal verbalization that makes such reflections possible” (p. 133). Nelson thus links language to inner speech, a process that the present reviewer finds central to self-awareness. Her analysis appears highly consistent with my own proposal (see my SCR paper on inner speech): “… a thought, a feeling, an experience from the past becomes ‘visible’ for examination by the thinking child when it is represented in verbal form” (p. 133).

In chapter 5 Marcel Kinsbourne discusses brain modules, evolution, zombies, and self-awareness. Like Nelson, he advocates a continuity/level approach to interspecific differences in self-reflection; he further states that “Any individual, human adult, child, animal, who can hold multiple representations in working memory can be self-conscious” (p. 155). This is a short theoretical contribution that I found refreshing, mainly because it looks at fish and bird behavior instead of primates.The fascinating chapter 6 by E. Tory Higgins is also theoretical and focuses on the motivational and evolutionary (adaptive) functions of self-reflective consciousness. Higgins starts by asking “How did self-knowing consciousness begin?” and readily admits that we don’t know. Instead, he prefers to ask “Once self-knowing consciousness did appear, why did it remain?” and answers “… because it serves basic self-regulatory functions… for getting along with others” (p. 157). More precisely, Higgins posits that “In order for children to obtain the nurturance and security they need to survive, they must establish and maintain relationships with caregivers who fulfill these needs… children must learn how their appearance and behaviors influence caregivers’ responses to them as objects in the world” (p. 158). To do so, children have to reflect on their selves and create a “self-digest”—a self-concept that ultimately includes a summary of what the world is like in relation to oneself. Once the self-digest has been formed, it helps satisfy two self-regulatory functions to self-awareness: shared reality and becoming. Shared reality refers to understanding what other people, with whom we share reality, hope for and expect of us; becoming means knowing where we are now, have been, and plan to become in relation to these hopes and expectations.Three other contributions by Janet Metcalfe and Hedy Kober (chapter 2), Herbert S. Terrace (chapter 3), and Patricia Kitcher (chapter 7) are fine, I’m sure, but they just did not stimulate my curiosity as much as other chapters in the book. Metcalfe and Kober introduce the notion of a “projectable self”, an entity that is self-reflectively conscious and that can “project” itself in the past and the future—Tulving’s concept of autonoetic consciousness. The authors review Any individual, human adult, child, animal, who can hold multiple representations in working memory can be self-conscious various types of evidence for self-awareness in human and nonhuman animals (self-recognition, Theory of Mind, deception, metacognition, and episodic memory). They conclude that “… there are some indications of self-reflective consciousness in some primates other than humans” (p. 77). Terrace examines how language affects metacognitive skills by taking into account such variables as vocabulary and grammar. And Kitcher, whose philosophical essay is a bit at odds with the other more empirically-oriented chapters, discusses animals’ capacity for episodic memory and metacognition is reference to Locke and Kant.

Conclusion

Overall I believe that Terrace and Metcalfe’s collection of papers on self-reflective consciousness represents an interesting and valuable addition to our current knowledge on higher-order mental processes in verbal and non-verbal organisms. As far as I can tell, the book summarizes most recent studies on episodic memory and metacognition in these organisms, very often in an engaging and clear style, sometimes in more complicated terms. Graduate students and researchers will find the book very helpful.

One final note about the title of the book itself. I find the title somewhat alluring and yet inaccurate. The “missing link” in cognition here obviously refers to self-reflective consciousness. As stated in my opening paragraph, and as explained throughout this review, the link still seems to be missing because the various manifestations of self-awareness used by contributors to measure it in nonhumans get overly simplistic. Then there is the “origins of self-reflective consciousness” part of the title, which may be misleading because only one chapter directly deals with this issue—Higgins’ brilliant contribution.

© 2006 Alain Morin

Author Information

Alain Morin, Ph.D
Behavioral Sciences, Mount Royal College, 4825 Mount Royal Gate S.W.,
Calgary, Alberta, Canada T3E 6K6
Email: amorin@mtroyal.ca
Web: http://www2.mtroyal.ab.ca/~amorin/