Signs and symbols: association or differentiation?
"Knowledge is fostered by curiosity; wisdom is fostered by awe. Awe precedes faith; it is the root of faith."
Much of my own research on nonhuman primates has been devoted to the problem of how the brain makes possible signs and symbols. For many years I questioned whether, in fact, nonhuman primates could construct signs and symbols but my doubts have now been resolved by work with two chimpanzees, one studied by the Gardners at the University of Nevada  and one by Premack at the University of California at Santa Barbara . The Nevada chimpanzee, named Washoe (after the county in which Renois located), has been taught to communicate using a sign language devised for the deaf and dumb. Earlier attempts to set up a rich communicative system between chimpanzee and man had failed. The Gardners felt that this failure was due to the limitations of the chimpanzee vocal apparatus and therefore decided to use a gestural system instead. The system chosen, American Sign Language, has the added feature that it is a relatively iconic rather than a phonetic system, thus much less complex in its structure than is human speech.
Washoe has learned to use approximately 150 signs. She can string two or three signs together but not in any regularly predictable order. Comparison with deaf human children of comparable age shows marked differences in the way in which gestural signs are used–but more of this later. The point here is that sign making is possible for the non-human primate.
The Santa Barbara chimpanzee, Sarah, is being trained by an entirely different method to an entirely different purpose. Premack has taken operant conditioning methods and applied them to determine just how complex a system of tokens can be used to guide Sarah's behavior. Experiments performed in the 1930's had already shown that chimpanzees will work for tokens–in fact a chimpomat had been constructed for use with poker chips. The chimpomat was an outgrowth of the delayed response task, the indirect form of which uses a temporary token to indicate where a piece of food (a reinforcer) is to be found subsequently. The delayed response task had been devised to determine whether animals and children could bridge a temporal gap between a momentary occurrence and a later response contingent on that occurrence. The bridge, which animals and children can construct, has been variously conceptualized in terms of "ideas," "memory traces," "short term memory organization," etc. Premack's chimpanzee has demonstrated that behavior dependent on tokens is not only possible but that hierarchical organizations of tokens can be responded to appropriately.
In all of these experiments the crux of the problem is that the token does not call forth a uniform response. Depending on the situation, that is, the context in which the token appears, the token must be apprehended, carried to another location, inserted into a machine or given to someone, traded for another token or traded in for a reward. Or, as in the original delayed response situation, the token stands for a reward which is to appear in one location at one time, another location at another time.
I shall use the term "symbols" to describe these context dependent types of tokens to differentiate them from "signs" which refer to events independent of the context in which they appear. (This distinction is consonant with that made by Chomsky, "Formal Properties of Grammars,"  and is used here to indicate that the primordia of the rules that govern human language are rooted in what are here called "significant" and "symbolic" processes.) There is now a large body of evidence to show that the cortex lying between the classical sensory projection areas in the posterior part of the brain is involved in behavior dependent on discriminating signs and that the frontal cortex lying anterior to the motor areas is involved in performances dependent on symbolic processes.
The surprise came when experiments were devised to show how these parts of the brain worked in determining sign and symbol. The ordinary view is that progressively more complex features are extracted or abstracted from information relayed to the projection areas: the simpler extractions occur in the projection areas per se, more complex abstractions demand relays beyond this primary cortex to adjacent stations where associations with information from additional sources (e.g. the primary projection areas) are made available . Unfortunately for this view there is a good deal of experimental evidence against it.
Most direct is the fact that if progressive cortico-cortical relays are involved in the ability to utilize signs and symbols, then removals of these relays should impair the ability. This is not the case. The posterior and frontal cortices specifically concerned in sign discrimination and in delayed response lie some distance from the primary sensory and motor areas. Complete removal of the tissue that separates the primary areas from those involved in discrimination and delayed response does not permanently impair the performance of these tasks: Fig. 4[20,47,48]. Ergo, cortico-cortical "abstractive" relays cannot be the mechanism at issue.
Two possibilities remain to explain the involvement of those cortical areas remote from the primary projection zones in discrimination and delayed response behavior. Information may reach these areas by routes independent of those that serve the primary projection cortex. This possibility is being actively explored in several laboratories. In the rhesus monkey, however, there is already evidence that these independent routes do not play the desired role: destruction of the pathways does not lead to a deficit in the performance of discriminations or delayed response [49,50].
The third possibility is one that I have been seriously exploring for the past decade and a half . This alternative holds that sign and symbol are constructed by a mechanism that originates in the cortex and operates on the classical projection systems in some subcortical location. Thus the effects of the functioning of the cortex involved in signing and symbolizing are conceived to be transmitted downstream to a locus where they can preprocess signals projected to the primary sensory and motor cortex. A good deal of evidence has accrued to this third alternative. Perhaps most important is the fact that a large portion of the pathway relays within the basal ganglia, motor structures of the motor mechanism of the brain:Fig 5. Sign and symbol manipulation thus involves the same brain structures that are used by the organism in the construction of acts. The suggestion that derives from these anatomical facts is that signifying and symbolizing are acts, albeit acts of a special sort.
There is, of course, a difference in the neuroanatomy involved in signifying and that involved in symbolizing. This difference, as well as the behavioral analysis of the tasks involved, tells a good deal of what these behavioral processes are all about. The pathways for signifying influence the primary sensory systems. Connections have been traced by electrophysiological techniques as far peripheral as the retina [53,54] and the cochlear nucleus , for instance. The connections important to the symbolic process have not as yet been determined as fully, but a good deal of the evidence points to involvement with the limbic systems structures on the innermost boundary of the forebrain .
This connection between limbic and frontal lobe function demands a word or two. Removal of tissue in these systems does not impair sign discrimination but does impair performance on such tasks as delayed alternation [57-59], discrimination reversal , shuttle-box-avoidance  and approach-avoidance, commonly called "passive" avoidance . In all of these tasks some conflict in response tendencies, conflict among sets, is at issue. The appropriate response is context (i.e. state) dependent and the context is varied as part of the problem presented to the organism. Thus a set of contexts must become internalized (i.e. become brain states) before the appropriate response can be made. Building sets of contexts depends on a memory mechanism that embodies self-referral, rehearsal or, technically speaking, the operation of sets of recursive functions. (The formal properties of memory systems of this type have been described fully by Quillian . The closed loop connectivity of the limbic systems has always been its anatomical hallmark and makes an ideal candidate as a mechanism for context dependency [64,65].
As an aside, it is worth noting that much social-emotional behavior is to a very great extent context dependent. This suggests that the importance of the limbic formations in emotional behavior stems not only from anatomical connectivity with hypothalamic and mesencephalic structures but also from its closed loop, self-referring circuitry. It remains to be shown (although some preliminary evidence is at hand [66,67] that the anterior frontal cortex functions in a corticofugal relation to limbic system signals much as the posterior cortex functions to preprocess sensory signals.
Thus signs and symbols are made by the brain's motor mechanism operating on two classes of images–in the case of signs those that encode sensory signals and in the case of symbols those that monitor various states of the central nervous system. Signs are codes invariant in their reference to events imaged–their meaning is context free. The meaning of symbols, on the other hand, is context dependent and varies with the momentary state induced in the brain by the stimulation. Both signs and symbols convey meaning, make possible a temporal extension of otherwise momentary occurrences.
Man shares the meaning conveyed by sign and symbol with nonhuman animals. This form of meaning, though perhaps more highly developed in man than in other animals, is not what makes him peculiarly human. Our search for man's unique thrust to make all his experiences and encounters meaningful needs to proceed to yet another level of complexity of encoding: only man makes propositions and reasons with them.