Influence of context: situated cognition

In secs. and we saw non-individualistic conceptions as a response to the dilemma of propositional attitudes. Need for a level of analysis higher than representational level (Marr and Dennett did not mention the cultural/ sociological level - sec. ).

In secs. and we mentioned situated cognition as a response to the problems thrown up by the classical symbolic-computation approach and thus as an example of a distributed/ bottom-up approach to cognition. Specifically, SitCog rejects symbolic AI's:

1. reliance on stored information and rules,
2. focus on conscious thought and
3. neglect of the social, cultural and physical context.

Philosophical arguments for situatedness were first made by Wittgenstein (Philosophical Investigations, 1958) and Heidegger (). In the Anthropology context by Geertz, + sociology, critical theory, political science, AI.

Linking of social, behavioural/psychological and neural perspectives of knowledge - providing a broader context to understand learning and cognition (W.J. Clancey, 1993 (Cognitive Science spl issue), 1997 (SitCog - book)).

Example of social/cultural perspective: ability to solve formally identical problems in mathematics or in naturalistic/ real-life contexts, eg. the Wason and Johnson-Laird problem.

Different strategies are available used in these two contexts. (Lave, Carraher) Carraher & Schliemann, 1982, 1983, studied fruit-vendors in Brazil. Found effective but idisyncratic strategies: eg. 12-year old coconut vendor with a Grade 3 education, asked him for 10 coconuts (price of one coconut was 35). The vendor's calculation: ``Three will be 105; with three more that will be 210. I need four more. That is ... 315 ... 350'' i.e. (3 x 35) + 105 + 105 + 35.

Lave (Cognition in Practice, 1988) ``... these ... studies ... challenge theoretical boundaries between activity and its setting, between cognitive, bodily, and social forms of activity, between information and value, [and] between problems and solutions'' (p. 3).

SitCog sees problem-solving strategies not as fixed attributes of either individuals or problems; it sees them emerge from an interaction between an individual's knowledge and the context in which that knowledge is to be applied ... implies a web of social and activity systems within which authentic practice takes shape.

```Situated'...does not imply that something is concrete and particular, or that it is not generalizable, or not imaginary. It implies that a given social practice is multiply interconnected with other aspects of ongoing social processes in activity systems at many levels of particularity and generality.''

- Lave, J. (1991). Situated Learning in Communities of Practice In L.B. Resnick, J.M. Levine and S.D. Teasley (Eds). Perspectives on Socially Shared Cognition Washington, DC: American Pscyhological Association. pp. 63-82.

Examples in Hutchins, E (1995) Cognition in the Wild Cambridge, Mass: MIT Press.

The symbolic-situated debate can be seen as a contemporary form of the rationalist-empiricist debate. Similarity of SitCog with behaviourism: avoid mental constructs, focus on environment and actions/ behaviour.

Two traditions of SitCog:

Type 1. Focus on collaboration, community participation, cultural construction (anthropologists, ethnographist, critical theorists). Vygotsky-inspired socio-historical school: Michael Cole, Jean Lave, Rogoff, Lucy Suchman.

Contrasts with the low-level responses dealt with by behaviourism, as also with Piaget's preoccupation with the individual and his/her interaction with the purely physical environment. ``Situated Action'' ``Situated Learning''

Type 2. Retains ip/symbolic AI's focus on individual mind and in that sense contrasts with behaviourism (also adds neural and physical embodiment - Clancy) In contrast to symbolic AI, this version of SitCog looks at lay knowledge rather than professional knowledge and puts practice before theory. (psychologist, linguists, generally AI-inspired cognitive scientists) Allan Collins, John Seeley Brown, Donald Norman, Bill Clancey ``Situated Cognition''

Theory	Data	Learning is	Loc. of structs	Unit of analysis
Behaviourism	Discrete events	skill acquisition	Envt	Individual
Symbolic Cog	Discrete info	concepts and strats	Mind	Individual
SitCog	Holistic context	effective participn	E+M	Individual (in type 1)
				Community (in type 2)

Problems with the situated approach:

1. Methodology is not well-specified.

2. How to explain or enable transfer?

3. Mistakes are ``fossilised.''