neurophysiology: an integration 



'953 



chain exists — of causality down levels and of purpose 

 up them (93, 97, no). 



MODELS 



A fitting culmination of a chapter such as this is an 

 examination of the conceptual models that have been 

 developed to understand the workings of the nervous 

 system and to predict its performance under various 

 circumstances. A great variety of these has been 

 offered, varying from direct inferences regarding 

 neuronal activity, based on neurophysiological 

 experiments, to formalistic considerations of hypo- 

 thetical entities in certain mathematical relations. 

 Most efforts have been primarily at the neuronal 

 level [e.g. Craik (49), Milner (212), Smithies (■_>(>(>) | 

 or primarily at the formal [e.g. Culbertson (51)] or 

 the philosophical level [e.g. Smithies (265)]; but the 

 most useful ones have attempted to develop the 

 mathematical relationships from the known properties 

 of neurons, using experiment. il values for the param- 

 eters of the equations [e.g. Rapoporl (24m, Harlow 

 (15)]. Most recently, models built upon communi- 

 cation engineering and information flow have joined 

 the parade [e.g. George (81)]. 



Certainly, with advanced neurophysiological 

 knowledge, and with the development and appli- 

 cation of appropriate mathematical tools, especiall) 

 those of relations rather than of quantity (and with 

 improved communication between those in the 

 separate fields and with effective interdisciplinary 

 training), models have increased in power and 

 sophistication. Some begin to have the sort of pre- 

 cision and elegance thai generate both confidence 

 in their congruence with reality and the emotional 

 satisfaction associated with achieving a real solution. 

 The physiological discoveries have all been in the 

 direction of giving greater freedom to the functioning 

 nervous system — freedom in time, in quantity, in 

 locus and in variability of performance- so that now 

 the pattern of activity is critically in focus, and 

 pattern is information or organization, the trade mark 

 of living organisms. Since adequately tracing the 

 development of models is out of the question, im- 

 portant advances will first be summarized and then 

 two of the more recent and satisfying models will be 

 given as examples. 



It need hardly be urged — except that some might 

 regard model building as 'mere speculation' (e.g. 

 Gernandt) — that theories or models enable man to 

 expand the possible world in his grasp, just as ex- 



periments enable him to restrict the possible toward 

 or, hopefully, to the actual. The problems todav are 

 to explore the consequences of various assumptions as 

 to the dynamics of groups of neurons, as multiple 

 properties range over permissible ranges. The use of 

 computers to simulate "brains' with assigned attributes 

 is beginning to pay off, as noted earlier; it is a rel- 

 atively cheap pretesting procedure to sharpen the 

 experiments with real brains. 



lit in h Marks "l .\iurajili\siology 



To present an evaluative summary of basic factual 

 and theoretic advances of the past centurv is rash; 

 but here is one nonetheless. 



vv>\ PROPERTIES. That nerves carried messages was 

 known to antiquity, but it was believed that they 

 served as passive conduits for hypothetical tluids or 

 vibrations. Over a century ago, the discovery of 

 resting and action potentials and the measurement of 

 conduction velocit) indicated the active partici- 

 pation of the nerve in conduction; and this was 

 full) borne out by the demonstration, in the first 

 half of this century, of the refractor) period, the 

 thermal and chemical changes associated with 

 activitv, and the impedance changes. Establishment 



of the all-or-none, or digital, behavior of the active 

 nerve fiber full) supported the core-conductor model 

 of a propagated activation. Depolarization of a 

 region activated it and dropped its resistance, thus 

 generating edd) currents which depolarized and so 

 activated a succeeding region. This model accounted 

 elegantl) lor .1 vast number of factual details, in- 

 cluding such influences as fiber diameter, myelin and 

 internode length on conduction velocity. It did not 

 account for the explosive membrane change and the 

 events leading up to it, nor did it pa) an) attention 

 to the delayed events following a single nerve action. 

 More careful examination of the membrane 

 potential and related properties revealed local 

 potentials that might oscillate or increment after a 

 brief shock, showed spontaneous threshold fluctu- 

 ations and, particularly, encountered the positive 

 overshoot of the action spike. The depolarization and 

 eddv currents were, therefore, also active responses ol 

 a living nerve rather than passive physical events; 

 indeed, for short neuron processes, electrotonic 

 currents ma) be the sole agent in activation spread. 

 The model that now emerged supplemented the 

 older one and related membrane resting potential to 

 the internal-external potassium ion ratio, the action 



