Book review of Peter Gardenfors

Peter Gardenfors:
HOW HOMO BECAME SAPIENS (Oxford Univ Press, 2003)

Home | The whole bibliography | My book on Consciousness

The Swedish linguist Peter Gardenfors has written an important book about the evolution of sentient life. He advances a thesis of how self-conscious beings came to be. First came sensations, then attention, then emotions, then memory, then thoughts (by which he really means "internal representations of the world"), then planning, then the self, then free will and finally language. Most of these faculties are not unique to humans. Most mammals have emotions and even thoughts. Chimpanzees exhibit all of these faculties up to planning. But Humans are the only animals that are truly conscious of themselves and can speak.
He makes one important assumption at the very beginning: that the "mind" (the cognitive being) is not limited to the brain but needs the entire organism. A brain without a body is not a mind. In fact, the entire environment (not only the organism's body) might be needed for a brain to work properly.
To justify his theory, he uses four sources of evidence: animal behavior, neurobiology, child development, archeology. Each of these provides some evidence about the cognitive faculties that are at play.
The cortex is the place where a representation of the world is created. That allows the brain to use the representation of an object (or a situation) rather than the object (or the situation) itself. It allows, in other words, to be somewhat "detached" from reality: the brain can work on something that is not an object/situation present "here and now". Gardenford believes that the large cortex of the human brain (i.e., its superior ability in representing the world) makes all the difference between human and animal behavior. Other animals have a cortex too, but it doesn't compare in size with the human cortex. Gardenfors believes that first came sensations, then perceptions (the interpretation of those sensations, which are already representations but are directly related to the world) and then "detached" representations (which he also calls "imaginations", and differ from perceptions which are "cued" representations, representations about something that is present here and now). Since all animals have sensations, Gardenfors assigns a degree of consciousness to all animals. But only mammals and birds have the cortex that allows for detached representations: they can "guess" and "plan". E.g., a cat does not need to see a mouse to understand that it is hiding in a place where it cannot be seen, and the cat can make a plan (by guessing how the mouse will behave) in order to catch it. Gardenfors explains the difference between sensations and perceptions as a difference in the referent: sensations are about what is happening to the body, whereas perceptions are about what is happening in the world (that is causing that change in the body). A perception is, in a sense, a step back to find out what caused the sensation. "We perceive the causes". Humans are better than any other animal at discovering the causes because they have better "simulators" in their cortex.
The next step up, the detached representations, are important because they can be used at any time, regardless of whether the object is present or not. They also provide an evolutionary advantage: the animal can play trial and error in its internal representation, without risking its life in the real world. The animal can simulate the consequences of acting before actually acting. An internal representation "allows our hypotheses to die instead of us". Animals that have an internal representation (which are animals with a large cortex) have some behavior in common: they play and they dream. Reptiles do not play and do not dream.
Thoughts (his nickname for "internal representations of the world") allowed some animals to "become increasingly detached from the immediate vicinity". Instead of reacting directly to stimuli from the environment, these animals can use "reason" to understand what is going on in the environment and to decide what to do next. An animal that can only react directly to a stimulus is limited to one course of action. An animal that can build an internal representation of the world is capable of creating more than one possible course of action.
The next step up is to actually "plan" an action. Many animals plan, but in an "immediate" fashion. Humans can plan in an "anticipatory" manner. The difference is about being ready for the same situation to occur again in the future. For example, animals make tools to be used immediately, but only humans carry their tools with them, knowing that they may need them again. Other animals would simply make the same tools again when required. Another example is how we communicate: animals do communicate, but their communication is about the "here and now", whereas humans can discuss of our memories of the past and dreams for the future. In a sense, another proof is the fact than only humans seem to be aware of the full meaning of death: they not only fear it, but are devastated by the mere thought of it (note that humans bury their dead, and this custom seems to be relatively recent in the evolution of humans).
Another ladder of cognitive abilities has to do with the kind of things that one's brain can represent: an internal representation of the world, which is necessary for immediate planning; "compassion" (an understanding of others' emotions); a theory of attention (understanding what others are focusing on); a theory of intention (understanding why others are doing what they are doing); a theory of others' minds (which is basically the ability to represent the internal representations of other minds); and finally self-consciousness (a representation of one's internal representation, which is required for anticipatory planning). It looks like the jump from understanding intentions and having a theory of others' minds is the difficult one: children acquire a theory of others at about the age of four; and it is still being debated whether apes ever do. Thus Gardenfors concludes that, in all likelihood, only humans are self-conscious.
The self, the last stage of human cognitive development, presupposes a you. Gardenfors assign a role even to deceit and cooperation. These are phenomena that presuppose an understanding of others' minds. The level of sophistication that the human race can achieve in matters of deceit and cooperation is due to the ability to work with the chain of nested beliefs: "I know", "I know that you know", "I know that you know that I know", etc. When one can see one's mind through the eyes of a competitor or a partner, one is seeing one's own mind. One can see one's own internal representation. Thus Gardenfors believes that an understanding of others' minds came before an understanding of one's own mind. I understand that you exist, act and have motives before I understand that I exist, act and have my own motives. First came the concept of "I and You", then came the concept of "I" (the subject, which presupposes a non-subject), and finally the concept of "it" (the object of the subject, which presupposes a subject).
Gardenfors believes that the self is an "emergent" phenomenon, a property of the whole that was not a property of any of its constituents. The "I" emerges from a network of ingter-related cognitive functions.
Gardenfors' theory of cognitive steps is consistent with Dennett's classification of "kinds of minds": "Darwinian creatures", which only live in the present; "Skinnerian creatures", which are capable of learning from trial and error; "Popperian creatures", which can play an action internally in a simulated environment before they perform it in the real environment; and "Gregorian creatures", which can extend their cognitive functions outside their organism by using tools and language.
A major departure of Gardenfors' theory from just about everybody else is the conclusion that language came last: not only was it unnecessary for the birth of consciousness, but consciousness is a primitive phenomenon and language is the last stage of cognitive evolution. Human language requires a kind of internal representation (the "detached" kind) that only humans have. Basically, it requires "symbols". In a sense, human language is about which is not here and not now, whereas other animals can only communicate about here and now, because their representations are not "detached" enough from external reality (they "are" about external reality). This limitation of other animals also translates in the sounds that they can produce. Humans are the only animals that can "choose" what sound to produce. Other animals have a repertory of sounds that they produce, and cannot control them. Humans can control them. Sometimes humans use the "instinctive" repertory of sounds (e.g., a scream or laughter). But humans can also articulate speech. Animals cannot literally talk. "They have no need to talk since they have nothing to talk about". They have no detached representation. They have no need to "talk" about things that are not here now. Gardenfors believes that even self-consciousness is required to be capable of speaking, because human language is very much about the "I" and the "you".
Gardenfors agrees with Robin Dunbar that, originally, language had a social function. Humans chatted for the same reason that apes groom each other: to cement social bonds. That helped humans create groups, and groups helped survive in a hostile environment. So language had an evolutionary advantage. Chomsky's theory of an innate universal grammar is unnecessary because grammar could be (yet another) emergent phenomenon that arises after speech already exists. The brain sort of organizes the speech acts that is performing. The result (not the cause) is the rules of grammar. Language is not handled by a separate "module" in the brain, as Chomsky claims, but is a natural evolution of cognitive skills that preexisted it. Gardenfors believes that the meanings of words are represented in and interpreted by the brain in manners similar to images (what is known as "cognitive semantics").
Gardenfors adds a fifth kind to Dennett's kinds of minds: "Donaldian" beings, named after Merlin Donald's third phase: Donald believes that about 50,000 years ago humans began to store memories in the outside world instead of in their own brain (e.g., cave paintings, figurines, calendars, etc). The invention of external memories (which does not imply any change in the structure of the brain) was fundamental for creating the kind of mind that we now have. Writing and science were simply further evolutions of that invention.
Ultimately, it is all about the internal representation, which in humans is "detached" enough to allow for thinking about the past and the future, and even for thinking about ourselves. Gardenfors sees evidence that humans have better "simulators" of the environment (building better representations) in apparently unrelated facts such as the human ability to aim and to beat a rhythm. Apes cannot aim and cannot keep time. The consequences of having good simulators are civilizations.