someofourfindings

Some of our findings

Our objective is quite simple, though difficult. We want to explain, in an integrated fashion:

(i) the recognition-primed decision model (think about what the nurse went through within minutes)

(ii) prospect theory and framing effects in complex economics and game-theoretical scenarios (think about Schelling's tax proposals' distinct framings)

(iii) complex cognitive decision-making, detached from external or political influences, such as chess playing (think about Capablanca's remark)

We are not bound to a specific methodology, so at any point in time you may find us conducting psychological experiments; developing computational architectures; doing some mathematical modelling; or writing up philosophical essays.

Here is a sampling of the basic philosophy underlying (and differentiating) our approach from that of most decision-scientists.

#1. When making decisions, humans do NOT choose between a large number of options

Indeed, humans do not even select between different alternatives! OK, this might sound a bit crazy at first, but there is a whole bunch of converging evidence for it. Our proposal is, instead, of 'quasi-binary' decisions.

More to come on this on an upcoming report named "the case for 'quasi-binary' decisions". In the meantime, the Minds and Machines paper gives some clues, though it does not present or discuss the converging evidence over this issue.

#2. It is NOT possible to cleanly separate perception from problem-solving from 'self-justification'

This makes it the case that if one wants to study decision-making or problem-solving, one must study perception.

See for instance the Schelling tax proposal; or Kahneman and Tversky's framing theories.

One of our studies pointing this out is: "A glimpse at the metaphysics of Bongard problems" published in Artificial Inteligence. See also the Hofstadter papers.

#3. Perception is about finding the essence of a situation

...and finding that essence consists of activating the right concepts in one's mind.

#4. Perception is achieved through analogy-making.

Humans have millions of information-gathering pathways, and all this information must converge to a useful form such as a number of active concepts. These concepts, we claim, are all-singing all-dancing entities according to the tune of the (contextual) moment, are brought by analogy-making, and structure complex scenarios into hierarchical chunks where items (things, objects, agents) fill some abstract roles.