|
RESEARCH IN SENSORY PROCESSES AS A MODEL FOR BEHAVIORAL SCIENCE
We have seen that in the sensory systems it is possible to describe output as a quantifiable function of input and to understand the function as a series of transformations each corresponding to a particular stage of the system. We have also seen that sensory information is selectively screened and combined so that the final output is more specific and contains a great deal of information. This process, achieved on the one hand by receptor specificities, and on the other hand by complex combinations of impulses through facilitation and inhibition of overlapping neural networks, enables an organism to respond to lines, moving objects, etc., rather than simpler and less important stimulus attributes. Will it ever be possible to quantify the stimulus-response relationship and understand its transformations in terms of neurophysiological processes for more complex phenomena such as, for example, love, sleep, dreaming, neurosis, language and aggression? There are at lest six immediately noticeable differences between the analysis of sensory processes and the analysis of behavior like that listed above. (1) The behavioral processes, their stimuli and responses appear to be more complex than sensory processes. Not only is the response more complex, involving many muscle and hormonal changes, but also the eliciting stimuli are more complex, consisting of many modalities and submodalities, and patterns of each. (2) The neural pathways are more complex. (3) The behavioral sequences appear to be more modifiable by learning. Thus we learn to love certain people, we learn to sleep on certain schedules, and we appear, to some extent at least, to learn a neurosis. Sensory analysis, as we have seen it, involves very little learning. (4) Single unit recordings, which have provided the breakthroughs in sensory physiology, must be done on anesthetized or otherwise immobilized animals, a condition which would seem to preclude research on many of the active behaviors above. (5) Sensory processes seem to be the same in animals as in man, allowing one to generalize from animal experiments, but many more complex behavioral sequences are known only from man and cannot be duplicated in animals. (6) Perhaps most important of all, complex behavior is difficult and perhaps impossible to quantify, both in terms of stimulus and in terms of response. (1) Complexity of stimulus and response. Complexity by itself should not daunt the scientist. The history of science shows time and again that mechanisms when unknown seem incredibly complex, but then when known, fall into simple patterns. Who could have foretold that the genetic mechanism would hinge on the order of four bases on a molecule? Who could have foretold that the complexities of speciation and the different types of animal life could be explained by one Darwinian theory? Or the properties of 96 chemical elements by eight positions on a periodic table? (2) Complexity of neural pathways. The sensory pathways are relatively easy to chart. They consist of isolated receptors, isolated afferent nerves and a few central pathways which have proved relatively easy to trace by evoked potentials. More complex behavior may turn out to involve more diffuse central pathways, glia as well as neuronal elements, DC fields as well as spike potentials, dendritic fields as well as parallel axons, all of which will make our understanding more difficult. (3) The role of learning. Here one must distinguish between content and process. The extent of one's dreams is in terms of his experience; the process presumably is universal. The person one falls in love with is determined by a unique combination of experiences; the process, presumably, is universal. And so forth. Content we can never understand fully. If the brain stores up a thousand bits of information per second and these are never erased, as Von Neumann estimates in The Computer and the Brain, then it would take an experimenter a lifetime just to sift through the experiences of one individual for a short time, even if he had full access to that experience! Process itself apparently changes to some extent as a result of learning, though we cannot even know this until we have closer analysis of process in one individual at one given time. Only when we have quantified a process in many individuals over a long period of time can we find out how much and why that process changes as a result of learning. Learning makes process more complex, but not impossible to understand. (4) the problem of recording. I do not know of any inherently insurmountable obstacles to the taking of single unit recordings from unrestrained, unanesthetized animals. Now, apparently it is technically unfeasible, though Sawa and Delgado and Hubel and Wiesel report single units from quiet, unanesthetized animals. but if there is one thing that we can learn from the history of science, that is that technical progress never stops. If a gadget can be built, it can be improved. One can foresee a day when we are getting single unit records from animals in the midst of normal activity. (5) the problem of generalization from animal to human. For many types of behavior, language, for example, this problem seems quite crucial. Yet as time goes on, animals are found to possess more and more human-like abilities. Dreaming has now been discovered in animals. Even language is not so far-fetched. Ferster is now training chimps to abstract numbers. Cathy Hayes describes teaching her baby chimp Viki to say rudimentary words. As for neurosis, it is difficult to lend much belief to the so-called "experimental neuroses" as duplicates of the human neuroses until we know more about the human variety, but as our knowledge of human neurosis proceeds, we can ask more and more relevant questions of the animal researchers, and conduct "experimental neuroses" on more and more relevant planes. As for mechanisms of sleep and aggression, there would appear to be little reason why we should not generalize now, especially if we analyze a genetic sequence leading up to man. If a particular behavior is found to develop through the primates on some sort of a predictable trend, then there is good reason to expect man to fit into that trend. (6) It is the faith of the neurophysiologist that the behavior of organisms ultimately may be explained by stimulus response "reflex arcs" although it may involve complex patterns including internal as well as external stimuli and may be modified by learning. But that is not to say that all behavior is quantifiable. For the sensory systems we can measure input to the eyes by a photoelectric cell, input to the ear by sound pressure level as recorded by a microphone, input to the olfactory and taste receptors by carefully weighing and mixing and applying a known concentration of test stimulus. And the outputs may be recorded quantitatively in terms of millivolt generator potentials or impulses per second in an afferent single fiber. Is there an analog to measure in more complex behavior? Or will it forever escape quantification? [There follows several paragraphs on quantified measurement of neurotic symptoms and sleep and dream phenomena.] It concludes that "if sensory physiology is to teach us about how to study the organism, these are the things it teaches: that input and output must be quantified and that one must quantify at various levels of the system. To summarize, it is my thesis that much of the complex behavior of organisms should be ultimately amenable to a quantified and mechanism-oriented understanding such as we are now gaining for sensory systems. The philosophy of this endeavor might best be derived from the Mullers and Helmholtzes and Adrians of sensory physiology, that the behavior of organisms is lawful and that these laws may be understood through the use of proper techniques. |