Creativity Research Journal 2007, Vol. 19, Nos. 2–3, 307–318

Copyright # 2007 by Lawrence Erlbaum Associates, Inc.

The Role of an Incubation Period in Creative Problem Solving Ut Na Sio Lancaster University, United Kingdom

Elisabeth Rudowicz Pomeranian Medical University, Poland

ABSTRACT: This experimental study tested the spreading-activation hypothesis that an incubation period helps to sensitize problem solvers to relevant concepts. The study also tested the selective forgetting hypothesis that an incubation period helps to desensitize problem solvers to irrelevant concepts. Chinese Chess GO players, 28 experts and 29 novices, solved 18 remote association tasks (RAT) and lexical decision tasks (LDTs) under immediate, rest, and incubation conditions. After each RAT, a set of LDTs incorporating the RAT solution and the irrelevant concept were presented, either immediately, or after a 2-min delay, or after a 2-min delay filled with incubation tasks. The findings of the study support the spreading activation hypothesis and suggest that spreading activation occurs only in a fixated mind. No support was found for the selective forgetting hypothesis.

Anecdotal reports (Ghislin, 1985; Wallas, 1926; Woodworth & Schlosberg, 1954) have documented that a flash of insight occurs after the problem solver puts the problem aside and shifts his or her attention to other areas for a while. This temporary abstention from the problem has been labeled the incubation period (Wallas, 1926). Several hypotheses have been proposed in an attempt to account for the incubation effect (Seifert, Meyer, Davidson, Patalano, & Yaniv, 1995). Two of the most prominent hypotheses are the spreading activation and selective forgetting hypotheses. The spreading activation hypothesis (Smith, 1995b; Yaniv & Meyer, 1987) proposes that during an

Creativity Research Journal

incubation period, activation spreads to the nodes representing the relevant concepts. Thus, problem solvers become more sensitive to them, and the problem solving process is facilitated. The selective forgetting hypothesis (Simon, 1966; Smith, 1995b; Smith & Blankenship, 1991) states that an incubation period provides time for suppressing the activation of the nodes representing the irrelevant concepts that fixate problem solvers’ minds, and in turn, problem solvers will become less sensitive to these irrelevant concepts. Thus, they can have a fresh view towards the problem and the problem solving process is facilitated. Experimental studies have been carried out to provide empirical support for the incubation effect. The setting of the early studies was, however, rather uniform. Namely, one group of participants was interrupted with a break (incubation period) while solving a creative problem, whereas the other group worked on the creative problem continuously. A number of studies have reported that the interrupted participants outperformed those working continuously (Fulgosi & Guilford, 1968, 1972; Mednick, Mednick, & Mednick, 1964; Murray & Denny, 1969), yet, none of them can offer support to the spreading activation or the selective forgetting hypotheses. This is because the superior performance may simply be the outcome of the reduction of mental fatigue or The procedures and the data reported in this article are part of a research project funded by the City University of Hong Kong. Correspondence should be sent to Ut Na Sio, Psychology Department, Lancaster University, Fylde College, Lancaster, LA1 4YF, United Kingdom. E-mail: [email protected]

307

U. N. Sio & E. Rudowicz

participants’ covert consciousness working on the problem during the incubation period. To correct this problem, ensuing incubation studies have modified the setting of the incubation. Instead of being unoccupied during the incubation period, participants had to engage in some intervening activities, thus preventing participants from working on the problem covertly or having any rest. These studies have shown that participants who had a filled incubation period performed better than those working on the problem continuously (Murray & Denny, 1969; Patrick, 1986; Penney, Godsell, Scott, & Balsom, 2004; Silveira, 1971) and than those who had a rest while solving the problem (Patrick, 1986; Penney et al., 2004; Smith & Blankenship, 1989). Moreover, participants performed better if they were exposed to a longer incubation period (Silveira, 1971; Penney et al., 2004). These findings not only offered substantial empirical support to the existence of an incubation effect, but also warranted that the performance improvement was not due to the mental fatigue dissipation or additional covert conscious work. However, it is still unclear whether the incubation effect is due to the occurrence of spreading activation, selective forgetting, or other unknown mechanisms. There are other studies claiming that they have adopted a more critical approach to examining the processes underlying an incubation. Dodds, Smith, and Ward (2002) and Dorfman (1990) examined the spreading activation hypothesis by investigating the effect on creative problem solving of presenting related cues during incubation. They argued that the spread of activation during an incubation should sensitize the problem solvers to related concepts, and thus they should be able to make use of the externally presented related cues to solve the problem. Yet, their argument may not have been correct, as one could argue assimilating external cues may also depend on how deep the problem solvers process the cues. In addition, these studies presented mixed results and cannot be accepted as strong evidence to support or reject the spreading activation hypothesis. The findings of studies testing the selective forgetting hypothesis are more convincing. Smith and Blankenship (1989, 1991) carried out a series of

308

experiments that yielded consistent results, indicating that an incubation period helped problem solvers increase their performance on creative problems. In these studies, participants had to solve a list of creative tasks, some of which were presented with irrelevant information. The presentation of irrelevant information fixated the participants’ minds and inhibited their performance. When solving the problems containing irrelevant information, participants performed better with an incubation period than if they had worked on the problem continuously. Also, a longer incubation period resulted in a higher performance and a larger degree of forgetfulness of the irrelevant concepts among the groups (Smith & Blankenship, 1989). This pattern of findings is consistent with the selective forgetting hypothesis that an incubation period can help to resolve the fixation effect and improve problem solvers’ performance by providing time for the irrelevant information to fade out. However, these findings cannot be generalized because, in these studies, the irrelevant information hindering the participants’ minds has been limited only to that externally encountered. In some situations, fixation may not be caused by the externally encountered stimuli, but can be internally generated by the automatic activation of the problem solver’s knowledge. A classic example of this type of fixation has been demonstrated in Duncker’s study (1945), which asked participants to fix a candle on a wall in a way preventing the wax from dripping onto the floor. They were presented with a box of matches and a few tacks. Participants could solve the problem only if they perceived the box as a platform rather than as a container. However, the participants’ traditional concept of the function of a box hindered their minds in perceiving the box as a platform. Another, more recent investigation demonstrating this type of fixation was Wiley’s (1998) study among baseball experts. Using Mednick’s (1962) method of remote association task (RAT), Wiley created RATs in which the first two stimuli were baseball-related, and the third stimulus and the solution were baseball-unrelated. The results revealed that baseball experts’ domain relevant knowledge would be activated automatically by the first two stimuli. Although the activated

Creativity Research Journal

Incubation Period

baseball knowledge was irrelevant to the tasks at hand, the baseball experts still could not ignore these internally generated irrelevant concepts. Thus, their performance was inhibited due to the fixation effect. Yet, the role of an incubation period in helping problem solvers to forget these kinds of irrelevant concepts has been not fully explained because the past incubation studies focused mainly on examining the role of an incubation period in desensitizing externally imposed irrelevant stimuli. The current study adopted a new approach to examine the spreading activation and selective forgetting hypotheses. First, instead of comparing participants’ creative problem-solving performance under different experimental conditions, this study used lexical decision task to directly assess participants’ sensitivity towards relevant and irrelevant concepts, before and after an incubation period. Conclusions from previous studies, showing that a person’s lexical decision time to a word is negatively correlated with the person’s sensitivity to that word (Meyer & Schvaneveldt, 1971; Scarborough, Cortese, & Scarborough, 1977), allow us to treat the lexical decision time as an indicator of a person’s sensitivity to these two types of concepts. Additionally, unlike the past incubation studies, the irrelevant concepts adopted in this study were automatically generated internally in the minds of the GO experts during the process of solving GO-misleading RATs. Based on the existing literature on the subject and with the focus of explaining the role of incubation period in creative problem solving, two hypotheses were formulated:

Method Participants Two groups of university students or recent graduates, from Hong Kong and Macau, were invited to participate in this study. All of them were native Cantonese speakers, and able to read and write traditional Chinese. The first group consisted of 28 GO experts, and the second group consisted of 29 GO novices. A GO expert was defined as a player who played GO regularly, whose total amount of time spent playing GO was 144 hr or above (equal to having 1 hr practice per week for at least 3 years), and whose GO ranking was above 10 kyu1. A GO novice was defined as a player whose total amount of time spent playing GO was less than 144 hr, and whose GO ranking was 10 kyu or below. Table 1 presents the participants’ age, gender2, major in university, GO ranking, and total amount of time spent playing GO for both the novice and expert groups. There was no statistically significant differences between these two groups regarding the gender, and field of study, (gender: X2 (1, N ¼ 57) ¼ 0.08, p ¼ 0.53; Field of Study: X2 (1, N ¼ 57) ¼ 0.18, p ¼ 0.41. However, the GO-experts were slightly older than the GO novices, p ¼ .007. Within each group, the age, GO ranking score, and the total amount of time spent playing GO among the three conditions were compared, and no statistically significant differences were detected. The respective p levels for the expert and novice groups were for age, p ¼ .31 and .44, GO ranking score, p ¼ .57 and 1

1. Spreading activation hypothesis: After an incubation period, both experts and novices will become more sensitive to concepts relevant to the unsolved RATs. Thus, the lexical decision time concerning these concepts will be shorter for both, the experts and novices. 2. Selective forgetting hypothesis: After an incubation period, experts will become less sensitive to the irrelevant domain-related concepts activated by the content of the RATs, whereas a similar effect will not be observed in novices. Thus, the experts’ lexical decision time concerning these concepts will be longer.

Creativity Research Journal

GO players have a rank that describes their playing strength. The ranking starts at 30 kyu (a total beginner) and goes down to 1 kyu as a player gets more skillful. After 1 kyu, a player becomes 1 dan, and from there the ranks ascend to 6 dan (the highest rank) 2 Although the gender distribution among the participants of the study was uneven (the majority were males), it is estimated that the influence of the gender imbalance on the results is negligible. This study examined the spread and inhibition of activation, occurring within GO-related and general knowledge networks. The imbalanced gender distribution could bias the measured data if there were any differences between males and females in the organization of the knowledge of these domains. However, gender differences in knowledge organization were identified only in the sexual and emotion-related domains (Manguno-Mire & Geer, 1998).

309

U. N. Sio & E. Rudowicz

Table 1. Breakdown of Age, Gender, GO Ranking, and Total Amount of Time Spent Playing GO (hours), Gender, Field of Study for Experts and Novices and for the Three Experimental Conditions Ranking Score"

Age Experimental Condition Expert Overall Immediate Rest Incubation Novice Overall Immediate Rest Incubation

Time

Gender

Field of Study

M

SD

M

SD

M

SD

Men

Women

Humanities

Science

Business

21.75 20.90 21.78 22.67

2.47 2.28 2.86 2.18

14.00 12.10 10.11 10.67

1.08 3.63 4.59 4.27

754.21 613.60 907.56 757.11

1037.74 669.96 1352.64 1117.45

24 8 8 8

4 2 1 1

5 2 1 2

2 0 2 0

21 8 6 7

20.41 20.08 20.57 20.86

1.32 0.37 0.43 0.50

35.17 30.15 33.22 34.75

1.07 8.47 6.67 2.44

57.10 65.85 59.11 38.29

43.54 49.55 39.49 35.69

23 12 6 5

6 1 3 2

8 3 2 3

4 0 2 2

17 10 5 2

"

Since there are two categories (kyu and dan) in the GO ranking system, a ranking score based on the participant’s GO ranking was computed for the sake of making comparisons between participants. The score ranged from 36 to 1, a score of 36 ¼ 30 kyu, a score of 16 ¼ 10 kyu, a score of 6 ¼ 1 dan, and a score of 1 ¼ 6 dan.

.36, and total amount of time spent in GO, p ¼ .84 and .41, respectively. Manipulation A Chinese language version of the RATs (Mednick, 1962) was created for the purpose of this study. The original RATs were developed to assess the problem solvers’ ability in making association between remote concepts. In each original RAT, three apparently unrelated words are presented to problem solvers, and problem solvers have to think of a fourth word that can form an association with each of the three words. For example, if the three stimulus words of a RAT are blue, rat, and cottage, the forth word can be cheese. There were three different types of RATs: neutral, GO-relevant, and GO-misleading. In the neutral RATs, the three words (stimuli) and the solution were GO-unrelated. In the GO-relevant RATs, the three words and the solution were GO-related. In the GO-misleading RATs, the first two words were GO-related, but the third word and the solution were GO-unrelated. The GO-related word that could form an association with the first two stimulus words was named the intrusive word. It was expected that the presentation of the first two stimulus words would mislead the experts to believe that the RAT was a GO-relevant one and that the intrusive word would be activated automatically. These intrusive

310

words were considered as the automatically activated irrelevant concepts. To confirm that the presentation of the first two stimuli would activate the expected intrusive word, a GO expert was invited to solve the GO-misleading RATs during a pilot study, and to write down the word that came to mind immediately after the presentation of the first two words. As predicted, the words written down by the GO expert were the expected intrusive words. The Appendix presents the Chinese RATs used in this study. Filler Tasks The mental rotation tasks (MRTs; Shepard & Metzler, 1971) and arithmetic tasks were adapted as filler tasks for occupying participants’ attention during the incubation period to prevent any conscious work on the RAT. Each of the 180 MRTs created for the purpose of this study consisted of a pair of three-dimensional objects. Each object consisted of 11 solid cubes attached face-to-face to form a rigid arm-like structure with exactly three right-angled elbows. The object placed on the left-hand side was the standard stimulus and the object on the right-hand side was the comparison stimulus. In half of the MRTs, the comparison stimulus was identical to the standard stimulus, whereas in the other half the comparison stimulus was the mirror image of the standard stimulus. The comparison stimulus was either rotated

Creativity Research Journal

Incubation Period

clockwise or counter-clockwise around the x-axis, y-axis, or z-axis with an angle. The rotation angle was 60, 80, 100, 120, 140, 160, or 180 degrees relative to the orientation of the standard stimulus. The arithmetic tasks included three-digit additions and subtractions, such as 127 þ 436 and 786 – 393.

and low in word complexity (with number of strokes less than 22). To minimize the unwanted semantic priming effect on the length of the lexical decision time, in each set of LDTs, every two consecutive words were assessed as being semantically unrelated by a native Cantonese speaker and could not form a meaningful two-word phrase in Chinese.

Dependent Variable The lexical decision tasks (LDTs) were adapted to examine the influence of an incubation period on the level of sensitivity to the two types of concepts, relevant to the unsolved RATs (solutions of the unsolved RATs) and irrelevant domain-related concepts (intrusive words of the unsolved GOisleading RATs). In the original LDTs, a word or pseudoword (letter strings that are pronounceable and look like a word, e.g, tuke or plone) was shown to participants, and they had to classify the item as a word or pseudo-word as quickly as possible. Their reaction time required for a lexical decision about a word was assumed as negatively correlated with their sensitivity to that word. In this study, a set of LDTs consisting of six items, including Chinese words and Chinese pseudowords, was presented after each RAT. The Chinese pseudowords were symbols composed of Chinese radicals arranged in wrong positions. If the RAT was a GO-misleading one, the six items were the RAT solution; the intrusive word; one, two, or three neutral words; and one, two, or three pseudowords. If the RAT was GO-relevant or neutral, the six items were the RAT solution, two or three neutral words, and two or three pseudowords. In order to assign the number of neutral and pseudowords to each LDT set, the number of neutral words was randomly selected first, and then the number of pseudowords was assign accordingly. In order to minimize the influence of word frequency and word complexity on the length of the lexical decision time, all Chinese words presented in the LDTs were high in word frequency (with occurrence of 100 or more in a million3) 3 The frequency count of the word was based on ‘‘A Chinese Talking Syllabary of the Cantonese Dialect: An Electronic Repository’’ (2002). Hong Kong: Chinese University of Hong Kong Humanities Research Institute. Retrieved October 8, 2004, from http://humanum.arts.cuhk.edu.hk/Lexis/Canton2/

Creativity Research Journal

Apparatus and Procedure Three experimental conditions, the Immediate, Rest, and Incubation, were created to test the two hypotheses. Participants were randomly assigned to complete the experimental tasks under one of the three experimental conditions. Table 1 presents participant’s distribution details and Figure 1 is the flow chart of the experiment. The experiment was run on a notebook computer using specially designed software. In the presentation of each RAT, participants had to write down their response on the Chinese writing pad connected to the computer. The time spent on solving each RAT, and the responses provided by participants were recorded by the computer. When solving the MRTs, the arithmetic tasks, and the LDTs, participants entered their responses via the keyboard. The reaction time and the responses given by the participants were recorded by the computer. The reaction time was measured from the onset of the stimulus until the participants responded. In a warm-up phase, participants in the Immediate and Rest Conditions were informed that the main experiment involved solving RATs and LDTs, and then they were prompted to solve three neutral RATs and three LDTs for practice. Participants in the Incubation Condition were informed that the main experiment involved solving RATs, LDTs, MRTs, and arithmetic tasks. They were then prompted to solve three neutral RATs and three LDTs, one MRT, and one arithmetic task for practice. After this warm-up phase, the main experiment started. During the main experiment, all participants were instructed to solve a sequence of RATs including six neutral RATs, six GO-relevant RATs, and six GO-misleading RATs organized in a random order. The order of the sequence

311

U. N. Sio & E. Rudowicz

Figure 1. The flow chart of the experiment.

presented to each participant was the same. In the presentation of each RAT, the first stimulus word was presented alone on the screen for 4 sec. Then the second stimulus word was placed next to the first stimulus word for 4 sec, followed by the third stimulus word, which was placed next to the second stimulus word. Participants had 30 sec to solve the RAT after the third stimulus word was presented. The participants could respond at any time during the 30-sec period by pressing the spacebar, writing down the response, and then pressing the spacebar again. Once the participants entered the answer, or the 30-sec period had passed, the participants were instructed to complete the corresponding set of LDTs. In the Immediate Condition, participants were prompted to do the LDT set immediately. In the Rest Condition, participants were prompted to sit quietly and listen to soft music for 2 min before proceeding to complete the LDT set. In the Incubation Condition, participants were prompted to complete five MRTs presented every 12 sec and four arithmetic tasks presented every 15 sec for the 2 min, before proceeding to complete the LDT set. This 2-min period occupied with mental activities was considered as an ‘‘incubation period’’. The inclusion of the Rest Condition group served the

312

purpose of assessing if the change in the speed of making lexical decisions was due to having a 2-min rest period before proceeding to complete the LDTs or due to the incubation period preceding the LDTs. The six items in the LDT set were presented on the screen one at a time. Once the participants responded, they were asked to press the spacebar to proceed to the next LDT. The serial positions of item types within each LDT set were spread evenly across the trial. An item in the first position of the sequence was always a neutral word or pseudoword, which served as a warm-up stimulus. The RAT solution and intrusive word occurred in the remaining five positions was mixed along with pseudowords and neutral words. They were requested to press the z key if the item represented a word and the slash key if the item represented a pseudoword. After the completion of the LDT set, the participants were requested to press the space bar in order to proceed to the next RAT (see Figure 1). Some of the RAT responses given by the participants did not match the expected solution, hence 12 university students were recruited to judge whether these unexpected responses were appropriate or not. Any response that was judged

Creativity Research Journal

Incubation Period

to be appropriate by more than six judges was categorized as a correct response. After the presentation of all experimental tasks, participants were asked to complete a paper-andpencil questionnaire. The questionnaire was aimed at collecting data on participants’ GO rankings, the number of years they have played GO, and the number of hr per week they spent playing GO. Participants were also asked to rate their degree of familiarity with the GO-related terms used in this study using a 6-point scale, ranging from 0 to 5 (0 ¼ not known, 1 ¼ known but very unfamiliar, 5 ¼ known and very familiar). This was a check to determine if all the participants knew all the GO-related terms used in this study.

Results Before conducting the main analysis, the experts’ and novices’ degree of familiarity with the GO-related terms was examined. All experts and novices rated at least 1 on their degree of familiarity with the GO terms, indicating that all participants knew the GO-related terms used in this study. But experts and novices differed in terms of their self-described depth of understanding of the GO-related terms used in this study. The mean familiarity rating of experts, M ¼ 4.62, SD ¼ .46, was higher than that of novices’, M ¼ 2.76, SD ¼ .77, t(55) ¼ 11.07, p < .001. To ensure that experts’ minds would be fixated by the automatically activated intrusive words when solving GO-misleading RATs, experts and novices’ performance on RATs were analyzed in terms of the number of correct responses. It was presumed that if the GO experts’ minds were fixated by the automatically activated intrusive words, their performance on solving the GO-misleading RAT should be inhibited, as compared with their performance on the other two types of RATs. A 2 (degree of expertise) $ 3 (RAT type) ANOVA on the number of correct responses was carried out to examine the experts’ and novices’ RAT performance. The interaction between degree of expertise and experimental condition was statistically significant, F(2, 50) ¼ 23.17, p < .01, hp2 ¼ .32. Subsequent pairwise comparisons using the LSD test revealed, for the expert group, that less GO-misleading RATs

Creativity Research Journal

Figure 2. Graph of the interaction between the Degree of Expertise and RAT Type on the number of correct responses on RATs.

(M ¼ 1.70, SD ¼ 1.23) were solved than both neutral RATs (M ¼ 2.70, SD ¼ 1.23, p < .01), and GO-relevant RATs (M ¼ 4.85, SD ¼ 1.17, p < .001). For novice group, there was no statistically significant difference in the number of correct responses among the three types of RATs: GO-relevant, M ¼ 2.41, SD ¼ 1.18; neutral, M ¼ 2.76, SD ¼ 1.27; GO-misleading, M ¼ 2.76, SD ¼ 1.75, GO-misleading versus neutral, p ¼ .15; GO-misleading versus GO-relevant, p ¼ .37. The experts’ and novices’ revealed performance patterns that were as predicted and are presented visually in Figure 2. Moreover, to check if participants in the three experimental conditions share a similar pattern of performance, two 2 (RAT type) $ 3 (experimental condition) ANOVAs, one for the novice group and one for the expert group, were carried out. The interaction effect between RAT type and experimental condition on the number of correct responses in both group was statistically nonsignificant; expert, F(4, 48) ¼ 0.54, p ¼ .71, hp2 ¼ 0.04; novice, F(4, 52) ¼ 2.057, p ¼ .10, hp2 ¼ 0.05; indicating that participants in the three experimental conditions share a similar pattern of performance. This implies that the predicted performance pattern was observed on participants in all the three experimental conditions and confirms that the GO-misleading RAT could induce a fixation effect on GO experts in the three experimental conditions. To verify the two hypotheses of this study, the sensitivity to the relevant (solutions of unsolved RATs), irrelevant (intrusive words of unsolved

313

U. N. Sio & E. Rudowicz

Table 2. The Means and Standards Deviations of the Original and Logarithmically Transformed (in bracket) Lexical Decision Time (in Milliseconds) Experts Source Solutions of the unsolved neutral RATs Immediate condition Rest condition Incubation condition Solutions of the unsolved GO-misleading RATs Immediate condition Rest condition Incubation condition Intrusive words of the unsolved GO-misleading RATs Immediate condition Rest condition Incubation condition Neutral words Immediate condition Rest condition Incubation condition

Novices

M

SD

M

SD

892 (2.91) 899 (2.90) 615 (2.77)

265 (0.13) 547 (0.20) 185 (0.10)

723 (2.82) 585 (2.75) 698 (2.80)

236 (0.13) 141 (0.09) 211 (0.09)

956 (2.94) 751 (2.84) 659 (2.80)

326 (0.14) 272 (0.14) 112 (0.07)

623(2.78) 627 (2.78) 722 (2.83)

163 (0.10) 144 (0.08) 214 (0.10)

1104 (3.00) 1024 (2.96) 745 (2.84)

396 (0.16) 327 (0.13) 153 (0.08)

779 (2.85) 750 (2.83) 795 (2.84)

316 (0.14) 252 (0.11) 270 (0.14)

1077 (2.97) 1010 (2.95) 872 (2.90)

309 (0.11) 254 (0.11) 98 (0.04)

789 (2.86) 750 (2.84) 797 (2.86)

134 (0.07) 164 (0.09) 158 (0.08)

Note. RAT ¼ Remote association task.

GO-misleading RATs), and neutral concepts (neutral words) was analyzed in terms of the participants’ length of lexical decision time for the words representing the relevant, irrelevant, and neutral concepts. However, the length of the lexical decision time for solutions of the unsolved GO-relevant RATs was excluded from the analysis due to the shortage of data from GO experts. This is because the majority of participants in the expert group (67.9%) solved at least five out of six GO-relevant RATs, and therefore insufficient data of the experts’ lexical decision time for the solution of unsolved GOrelevant RATs was collected. In the analysis, incorrect lexical decisions (4.68%) as well as correct lexical decisions with extremely long lexical decision times (longer than 2500 ms4) were discarded as outliers (5.80%). The lexical decision time was then logarithmically

4 Data analysis was carried out with cut-off points at 2000 ms, 2100 ms, 2200 ms, 2300 ms, 2400 ms, and 2500 ms. An extreme reaction time might be due to the factors unrelated to participants’ sensitivity to the stimulus, e.g., sudden shift of attention. In order to include as much data as possible in the analysis, the cut-off response time 2500 ms was picked as the outlier in this study.

314

transformed for diminishing skew (Keselman, Othman, Wilcox, & Fradette, 2004; Ratcliff, 1993). Table 2 presents the means and standard deviation of the original and transformed lexical decision time for solutions of the unsolved neutral RATs, solutions of the unsolved GO-misleading RATs, intrusive words of the unsolved GO-misleading RATs, and neutral words. The main focus of this study aimed at analysis of whether both novices and experts under the incubation condition will be more sensitive to relevant concepts, and experts under the incubation condition will be less sensitive to irrelevant concepts. Hence, a 2 (degree of expertise) $ 3 (experimental condition) MANOVA using the logarithmically transformed lexical decision time for these four types of words as four dependent variables was carried out to compare participants’ logarithmically transformed lexical decision time for these four types of words among the three experimental conditions. One MANOVA, rather than four separated ANOVAs, was carried out as it can better control the overall level of alpha (Field, 2000). The results of separate univariate analyses following the multivariate analysis were used to explore the effect of independent variables on each dependent variable. Table 3 presents

Creativity Research Journal

Incubation Period

the results of the univariate analyses in the MANOVA. The findings most relevant to the first hypothesis, concerning the effect of an incubation period on increasing sensitivity to relevant concepts, are the main effect of Experimental Condition on the length of the transformed lexical decision time for solutions of the unsolved neutral RATs and unsolved GO-misleading RATs. The main effect of Experimental Condition on the length of the transformed lexical decision time for both types of words was not statistically significant, revealing that participants in the incubation condition were not quicker than participants in other conditions in making lexical decision on these two types of words. The statistically nonsignificant findings may be a result of the incubation period influencing only a specific group (either experts or novices) and specific type of words (solution of unresolved GO-misleading RAT or solution of the unresolved neutral RAT). The statistically significant interaction effect of the degree of expertise and experimental condition on the length of the transformed lexical decision time for solutions of the unsolved GO-misleading RATs supports this explanation (see Table 3 and Figure 3). The subsequent pairwise comparisons using LSD indicated that in the expert group, the length of the transformed lexical decision time for solutions of the unsolved GO-misleading RATs under the

Incubation Condition was shorter than under the Immediate Condition, p ¼ .03. The results of similar pairwise comparisons made between Incubation and Rest Conditions, and between the Rest and Immediate Conditions, were not statistically significant (Incubation Condition vs. Rest Condition, p ¼ .54; Rest Condition vs. Immediate Condition, p ¼ .11). In the novice group, none of the results of pairwise comparisons made between the three experimental conditions were statistically significant (Incubation Condition vs. Rest Condition, p ¼ .72; Rest Condition vs. Immediate Condition, p ¼ .69; Immediate Condition vs. Incubation Condition, p ¼ .46]. The experts’ and novices’ performance patterns are presented graphically in Figure 3. The second hypothesis (selective-forgetting hypothesis) predicted that experts under the incubation condition would be less sensitive to the intrusive words of the unsolved GO-misleading RATs, hence, took a longer time to make lexical decisions on these words than experts under the other two conditions. However, experts under the incubation condition were slightly faster than experts under the other two experimental conditions in making lexical decisions on these types of words (see Table 2). Therefore, the second hypothesis was not supported by the findings of this study. A statistically significant main effect of expertise on the transformed response time to unresolved GO-misleading RATs, intrusive words, and neutral words was observed (Table 3), indicating that GO experts generally responded more slowly than GO novices in making lexical decisions.

Discussion

Figure 3. Graph of the interaction between the Degree of Expertise and Experimental Condition on the length of the lexical decision time for solutions of the unsolved GO-Mis leading RATs.

Creativity Research Journal

The results of this study revealed that the GO experts under the Incubation Condition took less time to make lexical decisions on solutions of the unsolved GO-misleading RATs than the GO experts under the Immediate Condition. This suggests that the GO experts are more sensitive to relevant concepts after an incubation period. One explanation of the increased sensitivity to relevant concepts is that activation spreads to relevant concepts during the incubation period to increase

315

U. N. Sio & E. Rudowicz

problem solvers’ sensitivity to them. It may be argued that the increase in lexical decision speed is simply due to the 2-min delay of doing LDTs. This argument can be rejected, as no statistically significant difference in the length of the transformed lexical decision time for the solution of unsolved GO-misleading RATs between the GO experts in the Rest and Immediate Conditions was observed. This pattern of findings also rules out the possibility that the increase in the sensitivity to the relevant concepts is the outcome of the mental fatigue reduction or covert conscious work on the problem during the 2-min break. If mental fatigue dissipation or covert conscious work is the underlying mechanism of the incubation period, the participants in the rest condition should benefit the most, as the unfilled 2-min break allows participants to refresh their minds or provide additional time for problem solving. However, the statistically nonsignificant difference between the Rest and the Immediate Conditions rejects this argument. There are also other alternative explanations to account for the decrease in lexical decision time. One of them is that the measured difference in the length of the lexical decision time for solutions of unsolved GO-misleading RATs is simply due to the difference in participants’ general lexical decision speed among the three experimental conditions. However, this possibility can be rejected as neither the main effect of Experimental Condition nor the interaction between degree of expertise and Experimental Condition on neutral words was statistically significant (see Table 3). It may still be argued that the increase in response time is not related to the increased sensitivity to the solution, but is facilitated by performing the incubation tasks, that is, the MRTs and arithmetic tasks. These tasks required participants to use the keyboard to enter answers and acted as a warm-up for the subsequent LDTs, which also required the use of a keyboard. This type of warm-up could increase participants’ speed in making lexical decisions, especially for participants who were slow in making lexical decisions. In this study, experts were generally slower than novices in making lexical decisions. Thus, experts might benefit more than novices from having an incubation period. This could account for

316

the findings that experts responded faster in making lexical decisions when solving unsolved GO-misleading RATs. If this interpretation is correct, then experts under the Incubation Condition should take a shorter time to make lexical decision on neutral words; however, the interaction effect was statistically significant only on the length of the transformed lexical decision time when solving unsolved GO-misleading RATs. Thus, this explanation can be rejected. In addition, in both expert and novice groups, participants’ age, GO ranking score, and total amount of time spent playing GO among the three conditions were comparable. This could guarantee that the measured difference in lexical decision tasks among the participants in the three experimental conditions were not the outcome of individual differences among them. On the whole, the pattern of the present findings provides strong and direct empirical support for the spreading activation hypothesis and the idea that spreading activation occurs during an incubation period to sensitize participants to the relevant concepts. It is worth noting that the observed incubation effect occurred only among the GO experts solving GO-misleading RATs. Since this type of domain-misleading problems can induce a fixation effect on GO experts, it seems that the incubation period does not always facilitate the creative problem solving, but only when the problem solvers’ mind is fixated. The relationship between the incubation period and the fixation effect suggested by this study is consistent with what Smith (1995a) proposed. Smith (1995a) argued that the incubation effect occurs primarily in situations in which initial fixation blocks a problem solver from solving a problem, and an incubation period helps to increase the problem solvers’ sensitivity to temporarily inaccessible relevant concepts when competing material causes fixation. The present findings, however, do not offer support to the selective forgetting hypothesis that the unconscious problem solving processes, occurring during an incubation period, suppresses the activation of the nodes representing the irrelevant concepts fixating the problem solvers’ minds. The GO experts examined in this study, after an incubation period, did not become less sensitive to the

Creativity Research Journal

Incubation Period

irrelevant domain-related concepts fixating their minds. The failure in revealing the desensitization to the internally generated irrelevant concepts during an incubation period may be due to the special nature of the irrelevant concepts and the task used for measuring the incubation effect. In this study, lexical decision tasks were used to assess participants’ sensitivity to the irrelevant GOrelated terms. GO-experts, who had extensive training in GO, were highly familiar with the GO-related terms, thus, the GO-experts were always responding to them with short lexical decision times. One could assume that even if the experimental incubation period helped the GO experts to become less sensitive to these irrelevant domain-related terms, this effect might still be too weak to overcome the effect of expertise and to increase the GO experts’ lexical decision time in response to them. It was somewhat surprising that the GO experts generally responded slower than GO novices in making lexical decisions (Table 2). One possible explanation of this observation is that the GO experts misperceived the lexical decision tasks as GO-related tasks. When solving domain-related tasks, experts tend to spend longer than novices at the beginning of the problem-solving process, examining the nature of the problem carefully, after which experts will execute the required problem solving steps more quickly. This problem solving style allows experts to outperform novices in terms of speed and accuracy. In this study, the GO experts may misbelieve that the LDT is a kind of GO-related task, and this may cause them to be more careful than novices at the beginning. Thus, this may lengthen GO-experts’ reaction times when making lexical decisions. Although this expert–novice difference is unexpected, it does not contradict the main findings of the study, as this study investigated the effect of an incubation period on the participants’ length of lexical decision time for relevant and irrelevant concepts, but not the incubation effect on general speed in making lexical decisions. On the whole, this study offers support to the spreading activation hypothesis, but no support to the selective forgetting hypothesis. In order to examine the role of an incubation period in desensitizing problem solvers to the internally generated

Creativity Research Journal

irrelevant concepts that fixate the participant’s mind, further modifications on the current experimental setting are needed. For example, researchers have to eliminate the biasing expertise effect by use other tasks, rather than the lexical decision tasks, to assess the degree of desensitization to the automatically activated concepts in the domain of their expertise. Besides this, the data of this study came from Hong Kong and Macau university students, and the task was specific only to one domain. Therefore, further studies have to be carried out to examine a broader range of participants (in terms of age, use of language, education level) and using a broader range of knowledge domains.

References Chinese University of Hong Kong Humanities Research Institute. (2002). A Chinese talking syllabary of the Cantonese dialect: An electronic repository. Retrieved October 8, 2004, from http://humanum.arts.cuhk.edu.hk/Lexis/Canton2/ Dodds, R. A., Smith, S. M., & Ward, T. B. (2002). The use of environmental clues during incubation. Creativity Research Journal, 14, 287–305. Dorfman, J. (1990). Metacognitions and incubation effects in insight problem solving. San Diego: Unpublished doctoral dissertation, University of California. Dunker, K. (1945). On problem solving. Psychological Monographs, 58. Field, A. (2000). Discovering statistics using SPSS for Windows: Advanced techniques for the beginner. London: Sage. Fulgosi, A., & Guilford, J. (1968). Short-term incubation in divergent production. American Journal of Psychology, 81, 241–246. Fulgosi, A., & Guilford, J. (1972). A further investigation of shortterm incubation. Acta Instituti Psychologici, 64–73, 67–70. Ghislin, B. (1985). The creative process. A symposium. Berkeley: University of California Press. Keselman, H.J., Othman, A. R., Wilcox, R. R., & Fradette, K. (2004). The new and improved two-sample t test. Psychological Science, 15(1), 47–51. Manguno-Mire, G. M., & Geer, J. H. (1998). Network knowledge organization: Do knowledge structures for sexual and emotional information reflect gender or sexual orientation? Sex Roles, 39(91–10), 705–729. Mednick, S. A. (1962). The associative basis of the creative process. Psychological Review, 69, 220–232. Mednick, M. T., Mednick, S. A., & Mednick, E. V. (1964). Incubation of creative performance and specific associative priming. Journal of Abnormal and Social Psychology, 69, 84–88. Meyer, D. E., & Schvaneveldt, R. W. (1971). Facilitation in recognizing pairs of words: Evidence of a dependence

317

U. N. Sio & E. Rudowicz

between retrieval operation. Journal of Experimental Psychology, 90(2), 227–234. Murray, H. G., & Denny, J. P. (1969). Interaction of ability level and interpolated activity (opportunity for incubation) in human problem solving. Psychological Reports, 24(1), 271– 276. Patrick, A. S. (1986). The role of ability in creative ‘‘incubation.’’ Personal and Individual Differences, 7(2), 169–174. Penney, C. G., Godsell, A., Scott, A., & Balsom, R. (2004). Problem variables that promote incubation effects. Journal of Creative Behavior, 38(1), 35–55. Ratcliff, R. (1993). Methods for dealing with reaction time outlines. Psychological Bulletin, 114, 510–532. Scarborough, D. L., Cortese, C., & Scarborough, H. S. (1977). Frequency and repetition effects in lexical memory. Journal of Experimental Psychology: Human, Perception and Performance, 3(1), 1–17. Seifert, M. C., Meyer, D. E., Davidson, N., Patalano, A. L., & Yaniv, I. (1995). Demystification of cognitive insight: Opportunistic assimilation and the prepared-mind perspective. In R. J. Sternberg & J. E. Davidson (Eds.), The nature of insight (pp. 65–124). Cambridge, MA: MIT Press. Shepard, R. N., & Metzler, J. (1971). Mental rotation of threedimensional objects. Science, 171, 710–713. Simon, H. A. (1966). Scientific discovery and the psychology of problem solving. In R. G. Colodny (Ed.), Mind and cosmos: Essays in contemporary science and philosophy (pp. 22– 40). Pittsburgh, PA: University of Pittsburgh Press.

Appendix.

Silveira, J. M. (1971). Incubation: The effect of interruption timing and length on problem solution and quality of problem processing. Dissertation Abstracts International, 32(09), 5500B. (UMI No. 7209560). Smith, S. M., & Blankenship, S. E. (1989). Incubation effects. Bulletin of the Psychonomic Society, 27(4), 311–314. Smith, S. M., & Blankenship, S. E. (1991). Incubation and the persistence of fixation in problem solving. American Journal of Psychology, 104, 61–87. Smith, S. M. (1995a). Fixation, incubation, and insight in memory and creative thinking. In S. M. Smith, T. B. Ward, & R. A. Finke (Eds.), The creative cognition approach (pp. 135–146). Cambridge, MA: MIT Press. Smith, S. M. (1995b). Getting into and out of mental ruts: A theory of fixation, incubation, and insight. In R. J. Sternberg & J. E. Davidson (Eds.), The nature of insight (pp. 65–124). Cambridge, MA: MIT Press. Wallas, G. (1926). The art of thought. London: Jonathan Cape. Wiley, J. (1998). Expertise as mental set: The effects of domain knowledge in creative problem solving. Memory & Cognition, 26(4), 716–730. Woodworth, R. S., & Schlosberg, H. (1954). Experimental psychology. New Delhi, India: Oxford & IBH. Yaniv, I., & Meyer, D. E. (1987). Activation and metacognition of inaccessible stored information: Potential bases for incubation effects in problem solving. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13(2), 187–205.

The List of Chinese RATs Created for This Study

Note. English translate in bracket.

318

Creativity Research Journal