Incubation processes during problem solving
To cite this output: Gilhooly, K. (2009). Incubation processes during problem solving: Full Research Report ESRC End of Award Report, RES-000-22-2686. Swindon: ESRC
REFERENCE No. RES-000-22-2686
Final Report to ESRC on grant RES-000-22-2686 August 28, 2009 K.J. Gilhooly & G. Georgiou
Incubation processes during problem solving Background Introduction Wallas (1926) proposed that an “Incubation” stage in problem solving, during which the problem is set aside and not consciously addressed, is beneficial. Dodds, Ward & Smith (2003) identified 39 relevant experiments of which 75% reported significant beneficial effects of incubation. More recently, Sio & Ormerod (2009) have reported an extensive meta-analysis which supports the beneficial effects of incubation in insight and divergent thinking tasks. The four main hypotheses regarding incubation effects that were addressed in the present project studies are as follows. 1. Intermittent Conscious work: This suggests that although incubation is intended to be a period without conscious work on the target task, nevertheless participants may carry out intermittent conscious work (Seifert, Meyer, Davidson, Patalano & Yaniv, 1995, p.82). Any conscious work during the supposed incubation period would reduce the time required when the target problem was re-addressed – but would impair performance on the interpolated task. As a check against this possibility, performance on the interpolated task during the incubation period should be compared with performance of a control group working on the same interpolated task without being in an incubation condition. A deficit in the interpolated task on the part of the incubation group would be consistent with the hypothesis of some conscious work on the target task during incubation. Although this seems a rather basic methodological check, surprisingly it has not been routinely carried out in previous research (Dodds et al., 2003). All the studies reported here incorporated suitable checks for conscious work on the target task during the incubation period. 2. Unconscious work: This approach argues that incubation effects occur through active but unconscious processing of the problem materials. Poincaré (1929) suggested that the “subliminal self” unconsciously combined and recombined ideas until an interesting relevant combination was formed whereupon the valuable idea would become conscious (i.e., Wallas’s Inspiration stage). More recently, Dijksterhuis and Meurs have applied a theory of unconscious thought (Dijksterhuis & Nordgren, 2006) to incubation. On this view, unconscious thought, compared to conscious thought, has a large capacity, proceeds relatively slowly, tends to be bottom up, is good at integrating many sources of information, is relatively poor at following rules and tends to be divergent rather than convergent. Dijksterhuis and Nordgren report a number of studies in which better decisions and better creative thinking were found when the tasks were not worked on consciously. Their studies did not follow the classical method of incubation research in which the problem is set aside after an extended period of conscious work. Rather, Dijksterhuis and Meurs had the participants immediately put aside the problem for a period after the task was presented, and before any conscious work could be carried out.
13
To cite this output: Gilhooly, K. (2009). Incubation processes during problem solving: Full Research Report ESRC End of Award Report, RES-000-22-2686. Swindon: ESRC
REFERENCE No. RES-000-22-2686
3. Selective Forgetting: This view (e.g., Simon, 1966) proposes an important role for automatic reduction in idea strength or activation. The proposal is that misleading strategies, mistaken assumptions and related “mental sets” weaken through forgetting and thus a fresh start or “set shifting” is facilitated when the problem is resumed. 4. Attention Withdrawal hypothesis: On this view (Segal, 2004), nothing happens during the incubation break. The break removes attention from a misleading assumption and on returning to the task there is a chance to “set shift” and adopt a more helpful assumption. This account is very close to the Selective Forgetting view; however, Segal argues that the two can be distinguished in that the Attention Withdrawal hypothesis predicts no effect of length of incubation interval as simple withdrawal of attention is sufficient on that hypothesis, while the Selective Forgetting view would suggest that longer incubation intervals which allow more forgetting would be more beneficial than shorter intervals. Segal (2004) reported a study with long and short incubation intervals in which both intervals were effective compared to no incubation interval but length of incubation interval was immaterial. This result is also counter to the Unconscious Work hypothesis according to which a longer incubation period would permit more unconscious work and hence more facilitation than would be found for a shorter interval. Unconscious work v. Attention Withdrawal: A current issue The hypotheses outlined above to explain incubation effects are not all mutually exclusive. Some of the hypothesised processes could co-exist and play roles within an individual problem solving episode. Different tasks may benefit from different processes and the exact positioning of the incubation period may affect which processes occur. However, among the hypotheses outlined above, a particularly stark contrast is offered by the Unconscious Work and the Attention Withdrawal hypotheses. Segal (2004), espousing the latter, argued that his results supported the view that nothing happens during incubation while Dijksterhuis and Meurs (2006) proposed that active unconscious thought processes must be invoked to explain their data. The conflicting conclusions of Segal (2004) and Dijksterhuis & Meurs (2006) may have resulted from a number of factors. One possibility is that the different conclusions are due to differences between the types of task used viz., a convergent spatial insight problem in Segal (2004) as against a divergent verbal task in Dijksterhuis and Meurs (2006). It is possible that Unconscious Work is helpful for divergent tasks where many alternatives are sought but not helpful for convergent tasks where only one solution path is possible. It seems a reasonable hypothesis that the positioning of the incubation period may also be significant. Incubation before any conscious work would minimise the role of forgetting (nothing yet to forget) or set switching (no sets established) and maximise any effects of Unconscious Work. The studies reported here investigated the explanations of the conflicting findings of Segal on the one hand and Dijksterhuis & Meurs on the other hand, by examining the effects of positioning of incubation opportunities on insight and divergent thinking tasks in different combinations from those represented by previous studies. It is notable that both Segal (2004) and Dijksterhuis and Meurs (2006) used interpolated tasks that were different in character from the target tasks. Segal’s target task was spatial while the interpolated tasks were verbal; Dijksterhuis and Meurs’s target task was verbal but the interpolated task was visuo-spatial. From Dodds et al.’s (2003) extensive review, the issue of similarity between target and interpolated tasks does not appear to have been addressed hitherto. The similarity relationship between target and interpolated tasks could be important in that different hypotheses suggest different effects of similarity. If the main process underlying incubation is beneficial forgetting then interpolated tasks similar to the
14
To cite this output: Gilhooly, K. (2009). Incubation processes during problem solving: Full Research Report ESRC End of Award Report, RES-000-22-2686. Swindon: ESRC
REFERENCE No. RES-000-22-2686
target task should promote more forgetting due to interference and hence bigger incubation benefits; however, if unconscious work is the main process then interpolated tasks similar to the target task should interfere with any unconscious work using the same mental resources and so lead to weaker (or even reversed) incubation effects when compared with effects of dissimilar interpolated tasks. Objectives 1. To fill a gap in the literature, in which insight problems had not been studied hitherto with incubation immediately after task presentation. 2. To examine effects of varying incubation durations and similarity of interpolated activities (verbal or spatial) with target task activities (verbal or spatial) in divergent and insight tasks. 3. To assess the theoretical implications of the results obtained. Methods and Results There were three experiments. Experiment 1 used a divergent task with delayed incubation and Experiments 2 and 3 used convergent insight tasks (one spatial and one verbal, respectively) with immediate incubation. Experiment 1. Effects of post-impasse incubation periods and congruence of interpolated activities on a divergent thinking task. In this experiment the target task was the divergent production of alternative uses for a brick. The positioning of the incubation periods was post-conscious work as used by Segal (2004) while the target task was as used by Dijksterhuis and Meurs (2006). Participants: A total N of 143 (106 female, 37 male) students at the University of Hertfordshire, with a mean age of 21.10 (SD = 3.89, range 18 – 40). Design: A 2 (incubation periods: 4 v. 8 mins.) X 2 (interpolated task: verbal v. spatial) independent groups design was used. The Ns per experimental group were as follows: 4 mins incubation and spatial interpolated task (N = 25); 4 mins incubation and verbal interpolated task (N = 23); 8 mins incubation and spatial interpolated task (N = 22); 8 mins incubation and verbal interpolated task (N = 26). There was also a control group (N = 47) that provided baseline performance data for target and interpolated tasks in the absence of incubation periods. Procedure: Participants were told that they would be asked to write down possible uses for a Brick different from the usual use. After 5 minutes working, participants were then assigned randomly to one of 2 incubation time periods (4 or 8 minutes) or to a control condition with no incubation period to continue for a further 2 minutes with no break. Participants assigned to the incubation conditions were told that they would be returning to the Brick Uses task later in the study. During the incubation periods participants either undertook verbal tasks (Anagrams) or spatial tasks (Mental Rotation items) presented in booklets. Sets of 73 five-letter single solution anagrams (from Gilhooly & Hay, 1977) and 48 mental rotation items (from Peters et al., 1995) were used and performance was scored in terms of numbers of items attempted and correct solutions during the period allowed. After the incubation periods there were further 2 minutes periods of work on the Uses task.
15
To cite this output: Gilhooly, K. (2009). Incubation processes during problem solving: Full Research Report ESRC End of Award Report, RES-000-22-2686. Swindon: ESRC
REFERENCE No. RES-000-22-2686
At the end of the Uses task participants were asked to indicate which of the uses they reported were subjectively novel i.e. had first occurred to them during the task rather than being previously known (following Gilhooly et al., 2007). The control group was tested to measure baseline performance on the intervening tasks (Anagrams and Mental Rotation) over 8 minute periods. From these data we also obtained baseline performance over a 4 minute period by marking progress every minute. These baseline measures were compared with performance on the tasks when used as intervening activities during the incubation periods. The same baseline data were used in our subsequent studies (rather than collect fresh baseline data for the same interpolated tasks repeatedly) which gave useful savings in costs and research time.
Results Effects of interpolation on the incubation period tasks Tables 1 and 2 below show performance on the rotation and anagram tasks when performed in control conditions for 4 or 8 mins and as interpolated tasks in the incubation conditions Table 1. Mental Rotation performance when carried out as an interpolated (incubation) task or as control task. Mean scores (max = 48) and SDs in brackets. Control Interpolated Task Time 4 mins 8 mins 4 mins 8 mins n items attempted 6.69 (2.65) 13.91 (4.54) 7.12 (3.14) 14.04 (5.60) n items correct 3.76 (2.58) 7.24 (4.53) 4.72 (2.89) 8.18 (4.43) From Table 1 it appears that carrying out Mental Rotation as an interpolated task during incubation periods did not affect Mental Rotation performance negatively and t-tests showed no significant differences at 4 minutes or 8 minutes interpolated v control performances in either number of items attempted or in number solved correctly. Table 2. Anagram performance when carried out as an interpolated (incubation) task or as control task. Mean scores (max = 73) and SDs in brackets. Control Interpolated Task Time 4 mins 8 mins 4 mins 8 mins n items attempted 14.28 (5.27) 26.56 (10.73) 18.09 (8.14) 28.38 (14.32) n items correct 12.58 (5.70) 19.43 (7.50) 14.43 (6.31) 20.78 (12.12) From Table 2 it appears that carrying out Anagrams as an interpolated task during incubation periods did not affect Anagram performance negatively and t-tests showed no significant differences at 8 minutes interpolated v control performances in either number of items attempted or in number solved correctly. At the 4 minute time period there was a significant increase in the number of anagrams attempted (t =2.24, df = 65, p
REFERENCE No. RES-000-22-2686
Final Report to ESRC on grant RES-000-22-2686 August 28, 2009 K.J. Gilhooly & G. Georgiou
Incubation processes during problem solving Background Introduction Wallas (1926) proposed that an “Incubation” stage in problem solving, during which the problem is set aside and not consciously addressed, is beneficial. Dodds, Ward & Smith (2003) identified 39 relevant experiments of which 75% reported significant beneficial effects of incubation. More recently, Sio & Ormerod (2009) have reported an extensive meta-analysis which supports the beneficial effects of incubation in insight and divergent thinking tasks. The four main hypotheses regarding incubation effects that were addressed in the present project studies are as follows. 1. Intermittent Conscious work: This suggests that although incubation is intended to be a period without conscious work on the target task, nevertheless participants may carry out intermittent conscious work (Seifert, Meyer, Davidson, Patalano & Yaniv, 1995, p.82). Any conscious work during the supposed incubation period would reduce the time required when the target problem was re-addressed – but would impair performance on the interpolated task. As a check against this possibility, performance on the interpolated task during the incubation period should be compared with performance of a control group working on the same interpolated task without being in an incubation condition. A deficit in the interpolated task on the part of the incubation group would be consistent with the hypothesis of some conscious work on the target task during incubation. Although this seems a rather basic methodological check, surprisingly it has not been routinely carried out in previous research (Dodds et al., 2003). All the studies reported here incorporated suitable checks for conscious work on the target task during the incubation period. 2. Unconscious work: This approach argues that incubation effects occur through active but unconscious processing of the problem materials. Poincaré (1929) suggested that the “subliminal self” unconsciously combined and recombined ideas until an interesting relevant combination was formed whereupon the valuable idea would become conscious (i.e., Wallas’s Inspiration stage). More recently, Dijksterhuis and Meurs have applied a theory of unconscious thought (Dijksterhuis & Nordgren, 2006) to incubation. On this view, unconscious thought, compared to conscious thought, has a large capacity, proceeds relatively slowly, tends to be bottom up, is good at integrating many sources of information, is relatively poor at following rules and tends to be divergent rather than convergent. Dijksterhuis and Nordgren report a number of studies in which better decisions and better creative thinking were found when the tasks were not worked on consciously. Their studies did not follow the classical method of incubation research in which the problem is set aside after an extended period of conscious work. Rather, Dijksterhuis and Meurs had the participants immediately put aside the problem for a period after the task was presented, and before any conscious work could be carried out.
13
To cite this output: Gilhooly, K. (2009). Incubation processes during problem solving: Full Research Report ESRC End of Award Report, RES-000-22-2686. Swindon: ESRC
REFERENCE No. RES-000-22-2686
3. Selective Forgetting: This view (e.g., Simon, 1966) proposes an important role for automatic reduction in idea strength or activation. The proposal is that misleading strategies, mistaken assumptions and related “mental sets” weaken through forgetting and thus a fresh start or “set shifting” is facilitated when the problem is resumed. 4. Attention Withdrawal hypothesis: On this view (Segal, 2004), nothing happens during the incubation break. The break removes attention from a misleading assumption and on returning to the task there is a chance to “set shift” and adopt a more helpful assumption. This account is very close to the Selective Forgetting view; however, Segal argues that the two can be distinguished in that the Attention Withdrawal hypothesis predicts no effect of length of incubation interval as simple withdrawal of attention is sufficient on that hypothesis, while the Selective Forgetting view would suggest that longer incubation intervals which allow more forgetting would be more beneficial than shorter intervals. Segal (2004) reported a study with long and short incubation intervals in which both intervals were effective compared to no incubation interval but length of incubation interval was immaterial. This result is also counter to the Unconscious Work hypothesis according to which a longer incubation period would permit more unconscious work and hence more facilitation than would be found for a shorter interval. Unconscious work v. Attention Withdrawal: A current issue The hypotheses outlined above to explain incubation effects are not all mutually exclusive. Some of the hypothesised processes could co-exist and play roles within an individual problem solving episode. Different tasks may benefit from different processes and the exact positioning of the incubation period may affect which processes occur. However, among the hypotheses outlined above, a particularly stark contrast is offered by the Unconscious Work and the Attention Withdrawal hypotheses. Segal (2004), espousing the latter, argued that his results supported the view that nothing happens during incubation while Dijksterhuis and Meurs (2006) proposed that active unconscious thought processes must be invoked to explain their data. The conflicting conclusions of Segal (2004) and Dijksterhuis & Meurs (2006) may have resulted from a number of factors. One possibility is that the different conclusions are due to differences between the types of task used viz., a convergent spatial insight problem in Segal (2004) as against a divergent verbal task in Dijksterhuis and Meurs (2006). It is possible that Unconscious Work is helpful for divergent tasks where many alternatives are sought but not helpful for convergent tasks where only one solution path is possible. It seems a reasonable hypothesis that the positioning of the incubation period may also be significant. Incubation before any conscious work would minimise the role of forgetting (nothing yet to forget) or set switching (no sets established) and maximise any effects of Unconscious Work. The studies reported here investigated the explanations of the conflicting findings of Segal on the one hand and Dijksterhuis & Meurs on the other hand, by examining the effects of positioning of incubation opportunities on insight and divergent thinking tasks in different combinations from those represented by previous studies. It is notable that both Segal (2004) and Dijksterhuis and Meurs (2006) used interpolated tasks that were different in character from the target tasks. Segal’s target task was spatial while the interpolated tasks were verbal; Dijksterhuis and Meurs’s target task was verbal but the interpolated task was visuo-spatial. From Dodds et al.’s (2003) extensive review, the issue of similarity between target and interpolated tasks does not appear to have been addressed hitherto. The similarity relationship between target and interpolated tasks could be important in that different hypotheses suggest different effects of similarity. If the main process underlying incubation is beneficial forgetting then interpolated tasks similar to the
14
To cite this output: Gilhooly, K. (2009). Incubation processes during problem solving: Full Research Report ESRC End of Award Report, RES-000-22-2686. Swindon: ESRC
REFERENCE No. RES-000-22-2686
target task should promote more forgetting due to interference and hence bigger incubation benefits; however, if unconscious work is the main process then interpolated tasks similar to the target task should interfere with any unconscious work using the same mental resources and so lead to weaker (or even reversed) incubation effects when compared with effects of dissimilar interpolated tasks. Objectives 1. To fill a gap in the literature, in which insight problems had not been studied hitherto with incubation immediately after task presentation. 2. To examine effects of varying incubation durations and similarity of interpolated activities (verbal or spatial) with target task activities (verbal or spatial) in divergent and insight tasks. 3. To assess the theoretical implications of the results obtained. Methods and Results There were three experiments. Experiment 1 used a divergent task with delayed incubation and Experiments 2 and 3 used convergent insight tasks (one spatial and one verbal, respectively) with immediate incubation. Experiment 1. Effects of post-impasse incubation periods and congruence of interpolated activities on a divergent thinking task. In this experiment the target task was the divergent production of alternative uses for a brick. The positioning of the incubation periods was post-conscious work as used by Segal (2004) while the target task was as used by Dijksterhuis and Meurs (2006). Participants: A total N of 143 (106 female, 37 male) students at the University of Hertfordshire, with a mean age of 21.10 (SD = 3.89, range 18 – 40). Design: A 2 (incubation periods: 4 v. 8 mins.) X 2 (interpolated task: verbal v. spatial) independent groups design was used. The Ns per experimental group were as follows: 4 mins incubation and spatial interpolated task (N = 25); 4 mins incubation and verbal interpolated task (N = 23); 8 mins incubation and spatial interpolated task (N = 22); 8 mins incubation and verbal interpolated task (N = 26). There was also a control group (N = 47) that provided baseline performance data for target and interpolated tasks in the absence of incubation periods. Procedure: Participants were told that they would be asked to write down possible uses for a Brick different from the usual use. After 5 minutes working, participants were then assigned randomly to one of 2 incubation time periods (4 or 8 minutes) or to a control condition with no incubation period to continue for a further 2 minutes with no break. Participants assigned to the incubation conditions were told that they would be returning to the Brick Uses task later in the study. During the incubation periods participants either undertook verbal tasks (Anagrams) or spatial tasks (Mental Rotation items) presented in booklets. Sets of 73 five-letter single solution anagrams (from Gilhooly & Hay, 1977) and 48 mental rotation items (from Peters et al., 1995) were used and performance was scored in terms of numbers of items attempted and correct solutions during the period allowed. After the incubation periods there were further 2 minutes periods of work on the Uses task.
15
To cite this output: Gilhooly, K. (2009). Incubation processes during problem solving: Full Research Report ESRC End of Award Report, RES-000-22-2686. Swindon: ESRC
REFERENCE No. RES-000-22-2686
At the end of the Uses task participants were asked to indicate which of the uses they reported were subjectively novel i.e. had first occurred to them during the task rather than being previously known (following Gilhooly et al., 2007). The control group was tested to measure baseline performance on the intervening tasks (Anagrams and Mental Rotation) over 8 minute periods. From these data we also obtained baseline performance over a 4 minute period by marking progress every minute. These baseline measures were compared with performance on the tasks when used as intervening activities during the incubation periods. The same baseline data were used in our subsequent studies (rather than collect fresh baseline data for the same interpolated tasks repeatedly) which gave useful savings in costs and research time.
Results Effects of interpolation on the incubation period tasks Tables 1 and 2 below show performance on the rotation and anagram tasks when performed in control conditions for 4 or 8 mins and as interpolated tasks in the incubation conditions Table 1. Mental Rotation performance when carried out as an interpolated (incubation) task or as control task. Mean scores (max = 48) and SDs in brackets. Control Interpolated Task Time 4 mins 8 mins 4 mins 8 mins n items attempted 6.69 (2.65) 13.91 (4.54) 7.12 (3.14) 14.04 (5.60) n items correct 3.76 (2.58) 7.24 (4.53) 4.72 (2.89) 8.18 (4.43) From Table 1 it appears that carrying out Mental Rotation as an interpolated task during incubation periods did not affect Mental Rotation performance negatively and t-tests showed no significant differences at 4 minutes or 8 minutes interpolated v control performances in either number of items attempted or in number solved correctly. Table 2. Anagram performance when carried out as an interpolated (incubation) task or as control task. Mean scores (max = 73) and SDs in brackets. Control Interpolated Task Time 4 mins 8 mins 4 mins 8 mins n items attempted 14.28 (5.27) 26.56 (10.73) 18.09 (8.14) 28.38 (14.32) n items correct 12.58 (5.70) 19.43 (7.50) 14.43 (6.31) 20.78 (12.12) From Table 2 it appears that carrying out Anagrams as an interpolated task during incubation periods did not affect Anagram performance negatively and t-tests showed no significant differences at 8 minutes interpolated v control performances in either number of items attempted or in number solved correctly. At the 4 minute time period there was a significant increase in the number of anagrams attempted (t =2.24, df = 65, p
