Page 404 - Deep Learning

Elements of a Unified Theory 387

the knowledge base which in turn determines the range of available options
and hence the probability of change. This, too, is under voluntary control.
Everyday discourse makes a sharp distinction between coping with change
and initiating change. But from a cognitive point of view, the key feature of either
case is the feedback that percolates through the system. Both positive and negative
feedback can alter the balance among options. in some cases, positive feedback
might elevate a previously unheeded option to win over its competitors. in others,
the habitual ways of thought and action might be so entrenched that the person
has to subject himself to negative feedback before change can occur. suppression
of past habits, whether of thought or action, has to be strong enough and frequent
enough that alternatives are given a chance to demonstrate their cognitive utility.
To suffer errors, failures, impasses and other difficulties is for some people an
experience to avoid, and the temptation is high to proceed where success comes
easy. But we can choose to ignore that temptation and expose ourselves to feed­
back in general and negative feedback in particular, and we can choose to con­
tinue our efforts in the face of repeated and persistent failures and impasses.
in short, choice of projects, level of engagement, time spent in preparation,
variability of experience and exposure to negative feedback are five factors that
affect the probability of non­monotonic changes and are subject to acts of will.
Different individuals make different choices in these respects for reasons that
have little to do with the nature of cognition and everything to do with their life
situation. Everything else being equal, the person who is born into a situation
in which others show little appreciation of, and no support for the intense work
required to build a vast knowledge base will be less prepared to undertake cre­
ative projects than someone at the opposite end of that dimension. The influ­
ence of life trajectory on the disposition for non­monotonic change is better
studied with the techniques of the biographer and the novelist than with those
of the cognitive psychologist. The important point for present purposes is that
differences among individuals in the disposition for non­monotonic change is,
in part, due to differences in life trajectories and hence do not necessarily imply
deep or intrinsic differences among their cognitive architectures.
indeed, the relation between variables like “flexibility” and a person’s record
of creative achievements, successful adaptations and belief conversions is likely
to be the opposite of the one commonly supposed: Every person responds to
some situations by projecting prior experience and to others by drawing back
to leap. When a leap results in a good outcome, we look back over the event
and say that the person behaved creatively, flexibly or with an open mind; if the
outcome was negative or unproductive, we describe the person in less flattering
ways. non­monotonic change is not caused by some mysterious power called