Developing Theories in Cognitive Psychology
Criteria for Differentiating between Different Scientific Theories
- Goodness of fit to empirical data
- Economy with assumptions (Occam's razor)
- Precision of formulation
- Sufficiency sufficient to account for the whole range of data regarding a phenomenon
- Completness - able to account for a complete range of behaviours
- Fruitfulness, whether right or wrong generates further developments
NB. The cognitive science criteria are apparently at odds with the traditional criteria
The Traditional Approach to Scientific Theory Construction
- Collect lots of data then think of low level theory to account for the data, the intermediate level theory to account for a set of low level theories, then eventually build the high level theory.
- e.g. Newton's Laws
- Theorists hoped that the information processing framework would provide necessary high level theory - but it seemed too limited - Said nothing about learning, semantic memory, the real world, affect etc. (Newell, 1973; Allport 1975; Neisser 1978)
- Conerns grew over the adequacy of the information processing framework and that Psychology theory was not building toward a high level theory, with too much prominence given to isolated low level phenomena that didn't build towards high level.
- Perhaps a combination of high and low level theories is needed
Criticisms of the H.I.P paradigm (see also Attention )
Newell A. (1973)
You can't play 20 questions with Nature and win. In W.G. Chase (Ed) Visual Information Processing.
(i) At the low level theories - Phenomenon-driven
- The recency effect
- The stroop pheneomenon
- The Muller-Lyer illusion
- The unattended speech effect
(ii) At the intermediate level theories - Dichotomy driven
- 1 vs 2 memory stores
- Nature vs Nurture "Dichotomies ofte have a habit of becoming points on a continuum, and so the theoretical issues, initaially clearcut, become muddier and muddier"
(iii) No progress towards high level theory
- [cf. Bakan (1967) research can be described as either 'bandwagon', 'leave no stone unturned', and 'fancy that']
- Allport (1975-BJP) also bemoaned the absence of cumulative progress.
General Critique of the Box Models of the 60s
- No clear understanding of form of input, output or process
(ii) Not sufficient
- Can't account for the range of known findings
- Don't generalise to other apects of cognition
- e.g. no model of semantic memory
(iv) Not falsifiable
- Very difficutl to devise direct tests
- Even if test failed, can always add more boxes
The Box Models of the 60's are an inadequate framework for understanding cognotion, however they may be a useful descriptive tool. Unfortunately, a much more detailed and comeplete model is needed before we can progress any further.