The Nature of Intuitive Thought by L. Järvilehto (2015)

© The Author(s) 2015
L. Järvilehto, The Nature and Function of Intuitive Thought and Decision Making,
SpringerBriefs in Well-Being and Quality of Life Research,
DOI 10.1007/978-3-319-18176-9_2

This is a book chapter I had a look at to think about what people mean when they talk about things being intuitive. I’ve been thinking a lot about how placement on a page or blackboard, writing style, colour etc. can make certain conclusions about the things being written feel intuitive, and what exactly that might mean. This chapter is a nice summary of the cognitive science on the topic.

There’s a lot of talk about System 1 and System 2, those being ‘in charge of autonomous and non-conscious cognition, and volitional and conscious cognition, respectively’ p. 23. These systems seem to be helpful way to think, albeit metaphorical:

The dual-system formulations of dual processing present a compelling picture of
how the mind works. As Evans and Frankish, among others, argue, these formulations
are, however, currently oversimplified. (Evans and Frankish 2009, p. vi).
According to Kahneman, the two systems are rather “characters in a story”—
abstractions used to make sense of how our cognition takes place. (Kahneman
2011, p. 19 ff.) He notes, “‘System 1 does X’ is a shortcut for ‘X occurs automatically.’
And ‘System 2 is mobilized to do Y’ is a shortcut for ‘arousal increases,
pupils dilate, attention is focused, and activity Y is performed.’” (Kahneman 2011,
p. 415). p. 28

Their relationship to working memory seems important:

One of the critical distinctions of the two types of processes is whether they
employ working memory. “In place of type 2 processes, we can talk of analytic
processes [that] are those which manipulate explicit representations through
working memory and exert conscious, volitional control on behavior” (Evans 2009,
p. 42). While the working memory is often likened to System 2, the two are not in
fact entirely the same:
Working memory does nothing on its own. It requires, at the very least, content. And this
content is supplied by a whole host of implicit cognitive systems. For example, the contents
of our consciousness include visual and other perceptual representations of the world,
extracted meanings of linguistic discourse, episodic memories, and retrieved beliefs of
relevance to the current context, and so on. So if there is a new mind, distinct from the old,
it does not operate entirely or even mostly by type 2 processes. On the contrary, it functions
mostly by type 1 processes. (Evans 2009, p. 37).
Type 2 processes need the constant application of working memory, such as in
calculating by using an algorithm, in evaluating various choices in decisionmaking,
or in practicing a new skill. p. 29

This is interesting because Cognitive Load Theory is based on working out how to reduce load on the working memory through design choices.

Engle points out, working memory is not just about memory, but rather using
attention to maintain or suppress information. He holds that working memory concerns
memory only indirectly, and that a greater capacity in working memory means a greater ability to control attention rather than a larger memory. (Engle 2002, p. 20.)

There’s a nice association between intuition and heuristics here, nice because relevance theory is so based in ideas about heuristics:

Gerd Gigerenzer presents a four-fold taxonomy for explaining intuitions.
According to Gigerenzer, gut feelings are produced by non-conscious rules of
thumb. These are, in turn, based on evolved capacities of the brain and environmental
Gut feelings are intuitions as experienced. They “appear quickly in consciousness,
we do not fully understand why we have them, but we are prepared to act on
them.” (Gigerenzer 2007, pp. 47–48.) The problem with the trustworthiness of gut
feelings is that many other things appear suddenly in our minds that bear a similar
clarity and that we feel like acting on, for example the urge to grab an extra dessert.
But not all such reactive System 1 behaviors are good for us.
Rules of thumb are, according to Gigerenzer, what produces gut feelings. These
are very simple heuristics that are triggered either by another thought or by an
environmental cue, for example the recognition heuristic, where a familiar brand
evokes positive feelings. (Gigerenzer 2007, pp. 47–48.) Evolved capacities are what
rules of thumb are constructed of. They include capacities such as the ability to
track objects or to recognize familiar brands. (Gigerenzer 2007, pp. 47–48.)
And finally, environmental structures determine whether a rule of thumb works
or not. The recognition heuristic may work well when picking up a can of soda or
even stocks, if it is directed towards trusted and well-known brands. (Gigerenzer
2007, pp. 47–48.) p. 41

Gary Klein has developed a similar position to Gigerenzer’s in his famous
decision-making research. In Klein’s recognition-primed decision making model,
decisions are made neither by a rational, conscious weighing scheme, nor by a fast
non-conscious calculation, but are based rather on quickly recognizing viable
strategies for action based on expertise. (Klein 1998.)
Like Gigerenzer’s, Klein’s idea is based on Herbert Simon’s conception of
intuition as recognition. According to Klein’s research, people do not in fact typically
make decisions by rationally evaluating choices. (Klein 1998, loc 202.)
Rather, a great majority pick up a choice that first comes to mind, mentally simulate
it, and if it seems to work, go with the first viable one, without ever considering
options. This decision-making scheme follows the strategy of satisficing, (accepting
the first viable option), made famous by Simon, in contrast to the more rational
strategy of optimizing, i.e. weighing all possible options and picking the one that
comes out on top as best. (Simon 1956.)
The difference between Gigerenzer’s and Klein’s positions is in that where
Gigerenzer assumes that gut feelings are produced by heuristics or rules of thumb
that are typical to all humans and produced by our environment, Klein’s idea of
recognition-priming is based on picking up much more individually complex
strategies of action based on prior experience and expertise.p. 42

The author works hard to distinguish ontogenetic from phylogenetic.

The gist here is that we generate a considerable amount of ontogenetic Type 1
processes, or habits, by exercise, deliberate practice and daily experience. p. 43

The author is also quite interested in situated mind ideas, and brings in questions of environment.

Martela and Saarinen delineate three principles of systems intelligence. First, we
must see our environment as a system we are embedded in. Second, we need to
understand that intelligent behavior cannot be traced back only to the capacities of
an individual, but arise as features of the entire system in which the individuals
operate. And lastly, intelligent behavior is always relative to a context. (Martela and
Saarinen 2008, p. 196 ff.) p. 48


A perceptual account of symbolic reasoning by David Landy, Colin Allen and Carlos Zednik

In what follows, we propose an  account of symbolic reasoning according to which perception, manipulation, and perceptual imagination lie at the heart
of mathematical and logical competence. p. 1

The authors reference Landy & Goldstone.

They talk about syntactic and computational accounts, and semantic processing accounts in why systems interpret and represent mathematical relations, including mental models, conceptual metaphors etc. This shifts the focus from syntax to meaningful relations. Both tend to ignore the perceptual role of the symbols themselves, even accounts that claim some grounded origin. p. 3

They contrast this with an Andy Clark cyborg account, a more fully situated cognition version in which the symbols actively help reasoning:

the active manipulation of physical notations plays the role of “guiding” the
human biological machinery through  an abstract mathematical
problem space—one that may far exceed the space of otherwise
solvable problems.p. 3

but point out that still not enough attention is paid to the effect on perception. They elaborate on the cyborg account, calling this Perceptual Manipulations Theory.


The neuralprocessesthatPMTtakestobeinvolvedinsym-
bolic reasoningalmostneverhaveastheirprimaryfunction
the implementationofamodally representedrulesormodels.
and perceptualorganization,objectrecognition,objecttrack-
ing andsymmetrydetection,amongothers.p. 4

They run through some examples of perceptual metaphors like the dy/dx one I wrote about so long ago.

¯A ∪ B = ¯A ∩ ¯B
¯P ∨ Q ≡ ¯P ∧ ¯Q

p. 5

They then outline some of the evidence that supports PMT, such as evidence that the physical form of notations significantly affects their interpretation/efficacy, etc.

A corollaryoftheclaimthatsymbolicandotherformsof
mathematical andlogicalreasoningaregroundedinawidevari-
tobebothidiosyncraticandcontext-specific. p. 7

overt rule-following emerges from the fine-
tuned interactions between the perceptual and sensorimotor
systems with well-designed physical notations—symbolic rea-
soning is a form of sophisticated “symbolpushing” that happens
to adhere to the formal rules of mathematics and logic,
due to a lengthy process of cultural adaptation and pedagogical
scaffolding. p. 8

The intelligent use of space by David Kirsh

Kirsh describes processes of organising a work space so as to pre-make decisions and make less on-line decision-making necessary; for example, laying out the vegetables to be cut in order clumps the decision-making at the beginning of the process, so that later as the other tasks begin all the cook has to do is follow the jig that she has laid out for herself. The possibilities open to her are restricted in such a way as to make the path that she earlier decided upon also the path that is easiest to follow without looking or thinking further.