One of the implicit, and sometimes explicit, objectives of modern neuroscience is to find neural correlates of subjective experience so that different qualities of that experience might be explained in detail by reference to the physical structure and processes of the brain. It is generally assumed that such explanations will make unnecessary or rule out any reference to conscious mental agents. This is the classic mind-brain reductivist program. We have chosen to challenge the optimism underlying such an approach in the context of sensory neurophysiology and sensory experience. Specifically, we ask if it is possible to explain the subjective differences among seeing, hearing, and feeling something by inspecting the structure and function of primary visual, auditory, and somesthetic cortex.

After reviewing the progress in localization of sensory functions over the past two centuries and examining some aspects of the structure and function of somesthetic, auditory, and visual cortex, we infer that one cannot explain the subjective differences between sensory modalities in terms of present day neuroscientific knowledge. Nor do present trends in research provide grounds for optimism.

At this point we turn to three philosophical theories to see what promise they hold of explaining these differences. A brief discussion of each – identity theory, functionalism, and eliminative materialism – suggests that none adequately accounts for the facts of the situation, and we tentatively conclude that some form of dualism is still a tenable hypothesis.

An individual has a theory of mind if he imputes mental states to himself and others. A system of inferences of this kind is properly viewed as a theory because such states are not directly observable, and the system can be used to make predictions about the behavior of others. As to the mental states the chimpanzee may infer, consider those inferred by our own species, for example, purpose or intention, as well as knowledge, belief, thinking, doubt, guessing, pretending, liking, and so forth. To determine whether or not the chimpanzee infers states of this kind, we showed an adult chimpanzee a series of videotaped scenes of a human actor struggling with a variety of problems. Some problems were simple, involving inaccessible food – bananas vertically or horizontally out of reach, behind a box, and so forth – as in the original Kohler problems; others were more complex, involving an actor unable to extricate himself from a locked cage, shivering because of a malfunctioning heater, or unable to play a phonograph because it was unplugged. With each videotape the chimpanzee was given several photographs, one a solution to the problem, such as a stick for the inaccessible bananas, a key for the locked up actor, a lit wick for the malfunctioning heater. The chimpanzee's consistent choice of the correct photographs can be understood by assuming that the animal recognized the videotape as representing a problem, understood the actor's purpose, and chose alternatives compatible with that purpose.

The term “stage” appears to be used in three general senses in theories of behavioral development: (a) as a metaphor: (b) as a description of behaviors that undergo age change; (c) as an explanation of age-related changes in behavior. Although most existing stage models are purely descriptive, a few of them purport to have explanatory power. One such model, Piaget's stages of cognitive development, is considered in this paper.

To be viewed as potentially explanatory, a stage model must describe some behaviors that undergo age change, posit antecedent variables believed to cause the changes, and provide procedures whereby the behavioral changes and the antecedent variables can be independently measured. Piaget's stages seem to satisfy some but not all of these requirements. Piaget's stages describe many agerelated changes in behavior, and some antecedent variables have been proposed. However, procedures do not exist for measuring the two factors independently. In lieu of such procedures, Piaget has outlined a “program” of five empirical criteria whereby the reality of his stages can ostensibly be verified. Some objections to these criteria are considered.

The five criteria in Piaget's program are invariant sequence, cognitive structure, integration, consolidation, and equilibration. Three of the criteria (invariant sequence, integration, and consolidation) lead to the same sorts of empirical predictions (culturally universal sequences in the acquisition of certain behaviors). Such predictions are subject to the objection that Piagetian invariant sequences are often measurement sequences. A measurement sequence is said to occur when some late-appearing behavior consists of some earlier-appearing behavior plus additional things. The cognitive structure criterion is subject to at least three criticisms: First, it yields, at most, descriptions of behavior; second, these are often nothing more than descriptions of task structure; third, they cannot be regarded as unique to the given stages for which they are posited. The fifth criterion, equilibration, generates some predictions that might be considered as prima facie evidence for the existence of stages. However, these predictions conflict with the current data base on Piaget's stages.

It is concluded that there is no compelling support for Piaget's hypothesis that his cognitive stages do more than describe age-related changes in behavior. Since explanatory statements involving stages appear with some regularity in Piagetian and neo-Piagetian writings, there are grounds for supposing this conclusion to be nontrivial.

Ethologists and comparative psychologists have found that the social behavior, discrimination learning, and especially the communicative behavior of many animals are sufficiently versatile to call into question the customary denial that animals have mental experiences comparable to our own. Many of their behavior patterns suggest that animals have mental images of objects, events, or relationships remote from the immediate stimulus situation, as well as self-awareness and intentions concerning future actions. Our behavioristic Zeitgeist has inhibited investigation of such possibilities, but reopening these long-neglected questions requires no departure from the materialistic approach to biological and behavioral science. The basic question at issue involves a comparison of those aspects of human and nonhuman brain function that sometimes lead to what we call mental experiences, at least in our own species. It seems more feasible to devise objectively verifiable tests for the existence of these general categories of mental experiences than for that of many others. While many other kinds of evidence could well be used (with appropriate caution and all possible controls) to weigh the likelihood that animals sometimes know what they are doing, the experimental use of communicative behavior seems especially promising, because the latter could serve a function analogous to human language, which provides our best evidence about the mental experiences of our fellow men.

The notion of the command cell has been highly influential in invertebrate neurobiology, and related notions have been increasingly used in research on the vertebrate nervous system. The term “command neuron” implies that the neuron has some critical function in the generation of a normally occurring behavior. Nevertheless, most authors either explicitly or implicitly use a strictly operational definition, independent of considerations of normal behavioral function. That is, command neurons are often defined as neurons which, when stimulated by the experimenter, evoke some behavioral response. Even when utilizing such an operational definition, investigators frequently differ on what they consider to be the exact characteristics that a neuron must have (or not have) to be considered a command cell. A few authors appear to treat command neurons in relation to normal function, but a precise behaviorally relevant definition has not been specified. Because of the ambiguity in the definition of command neurons, the term can refer to a wide variety of neurons which may play divergent behavioral roles. In some ways the attempt to label a cell as a command neuron may interfere with the process of discovering the complex causal determinants of behavior. Nevertheless, the notion that individual cells are responsible for certain behaviors is highly appealing, and an attempt to define the command neuron rigorously could be worthwhile. We suggest that a command neuron be defined as a neuron that is both necessary and sufficient for the initiation of a given behavior. These criteria can by tested by: (1) establishing the response pattern of the putative command neuron during presentation of a given stimulus and execution of a well specified behavior; (2) removing the neuron and showing that the response is no longer elicited by the stimulus (necessary condition); and (3) firing the neuron in its normal pattern and showing that the complete behavioral response occurs (sufficient condition). In some cases, groups of neurons, when treated as a whole, may satisfy the necessity and sufficiency criteria for a given behavior, even though individual neurons of the group fail to meet the criteria. We suggest that such a group be termed a “command system” for the behavior in question. Individual neurons in the command system can be termed “command elements” if, when fired in their normally occurring pattern, they elicit a part of the behavior, or “modulatory elements” if they do not in isolation elicit any response, but alter the behavior produced by other elements in the command system.

Two chimpanzees have demonstrated the ability to learn to use graphic symbols to ask one another for tools needed to obtain food. The chimps readily attended to and complied with one another's requests by selecting the appropriate tool from six distinct alternatives. They then shared the food obtained by this means and readily reversed the roles of tool requester and tool provider. Their joint accuracy was 92%, even when human experimenters were absent from the room. When they were prevented from using the graphic system (by deactivation of their keyboards), joint accuracy dropped to 10%, even though the animals could still see and hear one another and could gesture or vocalize freely. The acquisition of tool names revealed that words are at first closely and concretely linked to function, and hence that the traditional object-naming paradigm alone is a comparatively difficult starting point for the chimpanzee.

This work has demonstrated that two chimpanzees have been able to comprehend the symbolic and communicative function of the symbols they use. The results raise questions concerning chimpanzee research employing other kinds of symbolic communication systems. Apparently there is little evidence, apart from anecdote or uncontrolled tests, demonstrating that either Washoe or Sarah (chimps in other investigator's projects) has fully comprehended the symbolic nature and the communicative potential of their respective gestural and graphic token symbol systems.