Page 46 - Artificial Intelligence in the Age of Neural Networks and Brain Computing
P. 46

2. A Theoretical Method for Linking Brain to Mind  33




                  begin to understand the neural mechanisms and behavioral symptoms of mental
                  disorders on a continuum with neural mechanisms and behavioral properties of
                  typical behaviors.
                     Said in another way, after one does with due diligence discover and characterize
                  the brain mechanisms of normal behaviors, mechanistic explanations of clinical data
                  automatically emerge from these theories. In a similar way, the discovery of key brain
                  mechanisms, circuits, and architectures to explain one kind of data has often thrust me
                  into explanations of other, seemingly quite different, kinds of data where variations
                  and specialization of these mechanisms, circuits, and architectures are also operative.
                  In this sense, by getting the theoretical foundations of biological intelligence right,
                  one can then begin to reap the benefits of the gift that never stops giving.


                  2. A THEORETICAL METHOD FOR LINKING BRAIN TO MIND:
                     THE METHOD OF MINIMAL ANATOMIES
                  One cannot hope to derive a unified theory of an entire brain in one step, and one
                  should not try to do so. Rather, this grand goal can be achieved incrementally, in
                  stages, starting with a large behavioral database that excites a theorist’s imagination
                  (Fig. 2.1). The derivation begins with behavioral data because brain evolution needs
                  to achieve behavioral success. Starting with behavioral data enables models to be
                  derived whose brain mechanisms have been shaped during evolution by behavioral
                  success. Starting with a large database helps to rule out incorrect, but otherwise
                  seemingly plausible, models of how a brain works.
                     Such a derivation has always led in the past to the discovery of novel design prin-
                  ciples and mechanisms (Fig. 2.1) with which to explain how an individual, behaving





















                  FIGURE 2.1
                  A modeling method and cycle that clarifies how increasingly refined neural models can
                  explain and predict increasingly large interdisciplinary behavioral and neurobiological
                  databases.
   41   42   43   44   45   46   47   48   49   50   51