- Email: [email protected]
Commentary: Expertise and Extended Memory
Memory Performance in Experts
213
Commentary: Expertise and Extended Memory
K.J Gilhooly Aberdeen University, United Kingdom Ericsson and Pennington’s paper in this volume represents a welcome attempt to form bridges between the study of memory and that of problem solving. Too often these areas of study are pursued largely in isolation from one another. It is clear that in real ’everyday’ problem solving, memory processes must be implicated and equally that memory tasks in everyday life often have marked problem solving components. Expert problem solving has been extensively studied over the last 25-30 years and certain broad generalisations that seem to hold over a number of domains have been identified. Salient among these generalisations is the statement that experts remember new information in their fields better than do novices. This result was initially established in the area of chess by De Groot (1965) and has subsequently been replicated in a broad range of areas. As Ericsson and Pennington point out, such memory advantages are highly specific to the particular domains concerned and do not reflect any general superiority in basic memory capacities for experts. The specificity of such memory advantages goes against the view of expertise as a result of superior basic capacities (the ’Talent’ view). Rather, the memory results have been taken as support for the ’Knowledge’ view that expertise results from the accumulation of a large repertoire of domain-specific schemata which enable ready categorisation of problems into familiar types and support memory for fresh information in the domain. However as Ericsson and Pennington point out, sheer knowledge accumulation is unlikely to be the whole story and the development of memory skills are also very likely to be involved. Consideration of memory skills, as in Ericsson and Polson’s (1988a, 1988b) studies of waiters and Chase and Ericsson’s (1981; 1982) studies of practice effects in digit span tasks, has led to Skilled Memory Theory (Ericsson, 1985), which is applicable to expert memory performance also ( Ericsson & Kintsch, 1991). According to Skilled Memory Theory, skilled memory involves rapid storage of presented information in long-term memory via associations to existing information and retrieval via appropriate cues. Thus, in the digit span task subjects with extensive prior knowledge of running times associated chunks
274
Chapter 6
of digits with typical times for various distances and recalled via those distancetime links. It is suggested that experts have incidentally developed skilled memory for their domains. Experts have extensive background knowledge in long term memory to which new information can be readily linked; retrieval is by appropriate cues determined at time of encoding. Since it has been repeatedly shown that experts readily store and retrieve new domain information which has been externally presented it is plausible that during mental search in problem solving, self-generated information (e.g. on intermediate results/ evaluations etc.) is also rapidly stored in long term memory and readily retrieved as needed. One fruitful outcome of Ericsson and Pennington's analysis would be research directly bearing on expert memory for intermediately derived information during the course of problem solving. Experts should recall their intermediate steps and results more accurately and faster than should novices. Ericsson and Pennington's analysis is consistent with the traditional 'modal' model of human memory as consisting of a limited capacity short term or working memory backed up by an essentially unlimited capacity long term memory. What they have proposed is that experts circumvent the limitations of short term working memory by rapid storage in and retrieval from long term memory. Presumably, however, short term working memory is still a bottleneck in that information has to be retrieved into short term working memory in order to play a role in processing. I am assuming here that symbolic processing operates by the firing of rules that receive their inputs from short term working memory and place their outputs there also, the rules themselves being stored in long term memory (eg. the 'Process of Induction' model proposed by Holland, Holyoak, Nisbett & Thagard, 1986). Thus it seems worth distinguishing the extended working memory of readily available information in long term memory and the focal short term working memory. For reasons of cognitive control it would not seem desirable to suppose that the contents of extended working memory could activate rules in competition with the contents of focal short term working memory. Ericsson and Pennington appear to use the term 'working memory' in a basically functional way as referring to "internally held, readily available information". There is, of course, a different approach developed by Baddeley and colleagues which takes a structural view. On this approach working memory is regarded as a multicomponent system rather than as a homogeneous storage system. At present, three main components are distinguished, viz., a
Memory Petjormance in Experts
275
visuo-spatial subsystem, a verbal-phonological subsystem and a central executive subsystem for coordination and decision making. It is envisaged that each subsystem may require unpacking into component sub-subsystems (e.g. the visuo-spatial subsystem may require to be split into separate visual and spatial sub-subsystems). A key methodology has been the use of dual tasks held to load just one of the subsystems. So, for example, if a primary task can be carried out without decrement while the visuo-spatial subsystem is loaded by a secondary task then it can be inferred that the task does not involve the visuospatial subsystem (and vice versa). A considerable number of studies have supported the utility of this approach (see Baddeley, 1986; 1992; Logie, this volume for extensive reviews). How might the Baddeley et al. working memory model and Ericsson and Pennington’s extended working memory theory relate to one another? It could be argued that since Ericsson and Pennington’s ’extended working memory’ refers to long term memory use while Baddeley et al. are referring to focal short term working memory, the two groups are using similar terms but not in the same way and so there is no useful connection between the approaches. However, I suggest that Ericsson and Pennington are concerned with links between long term memory and focal short term working memory and that Baddeley et al.’s analyses and results are relevant to that linkage. As a specific example, if a task invokes a visuo-spatial strategy then information should be stored in that visuo-spatial form in the extended working memory as well as in the focal short term working memory in the interests of efficiency and compatability. Interestingly, there are some recent results bearing on this point. Saariluoma (1991) has carried out move-generating tasks in chess with concurrent secondary tasks aimed at loading the visuo-spatial or verbal components of working memory. The results indicated that visuo-spatial interference had the largest effect on rate of move generation, which implicates the visuo-spatial component of working memory in chess problem solving. A similar dual task study by Baddeley et al. (Baddeley, 1990, p.134) found that memory for briefly presented chess positions was most affected by a visuospatial secondary task (and by a central executive loading task) but not by a verbal subsystem loading task. Together these results suggest that visuo-spatially coded information is implicated both in the focal short term working memory during move-generation and in the extended working memory system which is tapped by the memory task.
216
Chapter 6
As a final comment, I would note that the extended working memory concept may be applicable to the processing of difficult problems at the edge of expertise. Such problems can tax scientists and artists for months or even years and any solutions are often regarded as creative. Helmholtz in 1896 (Woodworth & Schlosberg, 1954, p.838) noted that for successful solution of difficult problems "It was always necessary, first of all, that I should have turned my problem over on all sides to such an extent that I had all its angles and complexities 'in my head' and could run through them freely without writing." In other words, the complex problem representation had to be in extended working memory to overcome the limitations of focal short term working memory, before solution was possible. Thus, the notion of extended working memory looks promising in the analysis of extremely difficult problem solving as well as in cases of more routine expert problem solving.
References Baddeley, A. D. (1986). Working memory. Oxford: Oxford University Press. Baddeley, A. D. (1990). Human memory: Theory and practice. Hove, Sussex: Erlbaum. Baddeley, A. D. (1992). Is working memory working? Ouarterly Journal of Experimental Psychology, 44, 1-32. Chase, W. G. & Ericsson, K. A. (1981). Skilled memory. In J. R. Anderson (Ed.), Cognitive skills and their acquisition (pp. 141-189). Hillsdale, NJ: Erlbaum. Chase, W. G. & Ericsson, K. A. (1982). Skill and working memory. In G. H. Bower (Ed.), The psycholoev of learning and motivation, Vol. 16 (pp. 1-58). New York: Academic Press. De Groot, A. (1965). Thought and choice in chess. The Hague: Mouton. Ericsson, K. A. (1985). Memory skill. Canadian Journal of Psychology, 39, 188-231 . Ericsson, K. A., & Kintsch, W. (1991).
Memory in commehension and . (Report No.91-13) Boulder, Co: Institute of Cognitive Science, university of Colorado. p
G
Memory Performance in Experts
277
Ericsson, K. A. & Polson, P. G. (1988a). An experimental analysis of a memory skill for dinner orders. Journal of Experimental Psycholow Learning. Memory and Cognition, 14,305-316. Ericsson, K. A. & Polson, P. G. (1988b). Memory for restaurant orders. In M. Chi, R. Glaser, & M. Farr (Eds), The nature of exDertise (pp.23-70). Hillsdale, NJ: Erlbaum. Holland, J. H., Holyoak, K. J., Nisbett, R. E., & Thagard, P. R. (1986). Induction: Processes of inference, learning. and discovery. Cambridge, MA: MIT Press. Saariluoma, P. (1991). Visuo-spatial interference and apperception in chess. In R. H. Logie & M. Denis (Eds.), Mental images in human cognition (pp.83-94). Amsterdam: North-Holland. Woodworth, R. S . , & Schlosberg, H. (1954). Experimental psychology (3rd ed.). London: Methuen.
213
Commentary: Expertise and Extended Memory
K.J Gilhooly Aberdeen University, United Kingdom Ericsson and Pennington’s paper in this volume represents a welcome attempt to form bridges between the study of memory and that of problem solving. Too often these areas of study are pursued largely in isolation from one another. It is clear that in real ’everyday’ problem solving, memory processes must be implicated and equally that memory tasks in everyday life often have marked problem solving components. Expert problem solving has been extensively studied over the last 25-30 years and certain broad generalisations that seem to hold over a number of domains have been identified. Salient among these generalisations is the statement that experts remember new information in their fields better than do novices. This result was initially established in the area of chess by De Groot (1965) and has subsequently been replicated in a broad range of areas. As Ericsson and Pennington point out, such memory advantages are highly specific to the particular domains concerned and do not reflect any general superiority in basic memory capacities for experts. The specificity of such memory advantages goes against the view of expertise as a result of superior basic capacities (the ’Talent’ view). Rather, the memory results have been taken as support for the ’Knowledge’ view that expertise results from the accumulation of a large repertoire of domain-specific schemata which enable ready categorisation of problems into familiar types and support memory for fresh information in the domain. However as Ericsson and Pennington point out, sheer knowledge accumulation is unlikely to be the whole story and the development of memory skills are also very likely to be involved. Consideration of memory skills, as in Ericsson and Polson’s (1988a, 1988b) studies of waiters and Chase and Ericsson’s (1981; 1982) studies of practice effects in digit span tasks, has led to Skilled Memory Theory (Ericsson, 1985), which is applicable to expert memory performance also ( Ericsson & Kintsch, 1991). According to Skilled Memory Theory, skilled memory involves rapid storage of presented information in long-term memory via associations to existing information and retrieval via appropriate cues. Thus, in the digit span task subjects with extensive prior knowledge of running times associated chunks
274
Chapter 6
of digits with typical times for various distances and recalled via those distancetime links. It is suggested that experts have incidentally developed skilled memory for their domains. Experts have extensive background knowledge in long term memory to which new information can be readily linked; retrieval is by appropriate cues determined at time of encoding. Since it has been repeatedly shown that experts readily store and retrieve new domain information which has been externally presented it is plausible that during mental search in problem solving, self-generated information (e.g. on intermediate results/ evaluations etc.) is also rapidly stored in long term memory and readily retrieved as needed. One fruitful outcome of Ericsson and Pennington's analysis would be research directly bearing on expert memory for intermediately derived information during the course of problem solving. Experts should recall their intermediate steps and results more accurately and faster than should novices. Ericsson and Pennington's analysis is consistent with the traditional 'modal' model of human memory as consisting of a limited capacity short term or working memory backed up by an essentially unlimited capacity long term memory. What they have proposed is that experts circumvent the limitations of short term working memory by rapid storage in and retrieval from long term memory. Presumably, however, short term working memory is still a bottleneck in that information has to be retrieved into short term working memory in order to play a role in processing. I am assuming here that symbolic processing operates by the firing of rules that receive their inputs from short term working memory and place their outputs there also, the rules themselves being stored in long term memory (eg. the 'Process of Induction' model proposed by Holland, Holyoak, Nisbett & Thagard, 1986). Thus it seems worth distinguishing the extended working memory of readily available information in long term memory and the focal short term working memory. For reasons of cognitive control it would not seem desirable to suppose that the contents of extended working memory could activate rules in competition with the contents of focal short term working memory. Ericsson and Pennington appear to use the term 'working memory' in a basically functional way as referring to "internally held, readily available information". There is, of course, a different approach developed by Baddeley and colleagues which takes a structural view. On this approach working memory is regarded as a multicomponent system rather than as a homogeneous storage system. At present, three main components are distinguished, viz., a
Memory Petjormance in Experts
275
visuo-spatial subsystem, a verbal-phonological subsystem and a central executive subsystem for coordination and decision making. It is envisaged that each subsystem may require unpacking into component sub-subsystems (e.g. the visuo-spatial subsystem may require to be split into separate visual and spatial sub-subsystems). A key methodology has been the use of dual tasks held to load just one of the subsystems. So, for example, if a primary task can be carried out without decrement while the visuo-spatial subsystem is loaded by a secondary task then it can be inferred that the task does not involve the visuospatial subsystem (and vice versa). A considerable number of studies have supported the utility of this approach (see Baddeley, 1986; 1992; Logie, this volume for extensive reviews). How might the Baddeley et al. working memory model and Ericsson and Pennington’s extended working memory theory relate to one another? It could be argued that since Ericsson and Pennington’s ’extended working memory’ refers to long term memory use while Baddeley et al. are referring to focal short term working memory, the two groups are using similar terms but not in the same way and so there is no useful connection between the approaches. However, I suggest that Ericsson and Pennington are concerned with links between long term memory and focal short term working memory and that Baddeley et al.’s analyses and results are relevant to that linkage. As a specific example, if a task invokes a visuo-spatial strategy then information should be stored in that visuo-spatial form in the extended working memory as well as in the focal short term working memory in the interests of efficiency and compatability. Interestingly, there are some recent results bearing on this point. Saariluoma (1991) has carried out move-generating tasks in chess with concurrent secondary tasks aimed at loading the visuo-spatial or verbal components of working memory. The results indicated that visuo-spatial interference had the largest effect on rate of move generation, which implicates the visuo-spatial component of working memory in chess problem solving. A similar dual task study by Baddeley et al. (Baddeley, 1990, p.134) found that memory for briefly presented chess positions was most affected by a visuospatial secondary task (and by a central executive loading task) but not by a verbal subsystem loading task. Together these results suggest that visuo-spatially coded information is implicated both in the focal short term working memory during move-generation and in the extended working memory system which is tapped by the memory task.
216
Chapter 6
As a final comment, I would note that the extended working memory concept may be applicable to the processing of difficult problems at the edge of expertise. Such problems can tax scientists and artists for months or even years and any solutions are often regarded as creative. Helmholtz in 1896 (Woodworth & Schlosberg, 1954, p.838) noted that for successful solution of difficult problems "It was always necessary, first of all, that I should have turned my problem over on all sides to such an extent that I had all its angles and complexities 'in my head' and could run through them freely without writing." In other words, the complex problem representation had to be in extended working memory to overcome the limitations of focal short term working memory, before solution was possible. Thus, the notion of extended working memory looks promising in the analysis of extremely difficult problem solving as well as in cases of more routine expert problem solving.
References Baddeley, A. D. (1986). Working memory. Oxford: Oxford University Press. Baddeley, A. D. (1990). Human memory: Theory and practice. Hove, Sussex: Erlbaum. Baddeley, A. D. (1992). Is working memory working? Ouarterly Journal of Experimental Psychology, 44, 1-32. Chase, W. G. & Ericsson, K. A. (1981). Skilled memory. In J. R. Anderson (Ed.), Cognitive skills and their acquisition (pp. 141-189). Hillsdale, NJ: Erlbaum. Chase, W. G. & Ericsson, K. A. (1982). Skill and working memory. In G. H. Bower (Ed.), The psycholoev of learning and motivation, Vol. 16 (pp. 1-58). New York: Academic Press. De Groot, A. (1965). Thought and choice in chess. The Hague: Mouton. Ericsson, K. A. (1985). Memory skill. Canadian Journal of Psychology, 39, 188-231 . Ericsson, K. A., & Kintsch, W. (1991).
Memory in commehension and . (Report No.91-13) Boulder, Co: Institute of Cognitive Science, university of Colorado. p
G
Memory Performance in Experts
277
Ericsson, K. A. & Polson, P. G. (1988a). An experimental analysis of a memory skill for dinner orders. Journal of Experimental Psycholow Learning. Memory and Cognition, 14,305-316. Ericsson, K. A. & Polson, P. G. (1988b). Memory for restaurant orders. In M. Chi, R. Glaser, & M. Farr (Eds), The nature of exDertise (pp.23-70). Hillsdale, NJ: Erlbaum. Holland, J. H., Holyoak, K. J., Nisbett, R. E., & Thagard, P. R. (1986). Induction: Processes of inference, learning. and discovery. Cambridge, MA: MIT Press. Saariluoma, P. (1991). Visuo-spatial interference and apperception in chess. In R. H. Logie & M. Denis (Eds.), Mental images in human cognition (pp.83-94). Amsterdam: North-Holland. Woodworth, R. S . , & Schlosberg, H. (1954). Experimental psychology (3rd ed.). London: Methuen.