Page 141 - Deep Learning
P. 141
124 Creativity
explain why people fail to make progress on a problem that they are capable
of solving. In his 1935 monograph Zur Psychologie des productiven Denkens,
Karl Duncker, one of the Gestalt psychologists, advanced the idea that famil-
iar objects are encoded in memory in terms of their common functions, as
54
opposed to (or in addition to) their physical attributes. A hammer is encoded
as something to strike a nail with as well as something that is heavy, long and
metallic. When we perceive an object, we automatically retrieve from mem-
ory its familiar function and we spontaneously think about it in terms of that
function. This hypothesis became entrenched in the English research literature
as functional fixedness. The reality of functional fixedness is not in doubt; it
55
can easily be produced in the laboratory. But many problems that produce
unwarranted impasses do not involve the use of any objects, familiar or other-
wise, or do not require that those objects are used in any novel way. In solving
the Nine-Dot Problem the person uses a pen or pencil in its standard function
of drawing lines on paper. Functional fixedness is not an alternative explana-
tion of unwarranted impasses, but merely a special case of the general princi-
ple that impasses are caused by the activation of unhelpful prior knowledge,
applicable only when the relevant prior knowledge pertains to the common
uses of familiar objects.
Abraham H. Luchins demonstrated that training can make a person blind
56
to possible solutions. In the Water Jar Problem, a person starts with a water
supply and three unmarked beakers of known volume, and the goal is to end
57
up with a certain amount of water in the largest beaker. Luchins showed that
if subjects are given a sequence of problems that require a long solution S, and
then a problem that can be solved by the long solution and also by a shorter
solution Sʹ, they will as a rule overlook the shorter solution. This effect has
entered psychology under the German term Einstellung; the English word set
has a very similar meaning. Like functional fixedness, Einstellung is merely a
special case of the general principle that activation of unhelpful biases – in this
case, biases acquired as a result of training deliberately contrived to mislead –
causes unwarranted impasses.
More recent rediscoveries of this principle include the concept of a mental
rut and the constraining effects of expertise on problem solving. In research
58
on skill acquisition, the same idea is discussed under the terms automatic-
ity, capture errors, negative transfer, negative priming and task switch costs, as
if every generation and every group of researchers feel a need to make the
phenomenon their own by giving it a new name. Because the data gathered to
document the specific cases lend support to the general principle, the latter has
a broad basis and can be considered established.