CYBERPSYCHOLOGY & BEHAVIOR Volume 12, Number 1, 2009 © Mary Ann Liebert, Inc. DOI: 10.1089/cpb.2008.0147

Dissociation of Past and Present Experience in Problem Solving Using a Virtual Environment Bradley R. Sturz, Ph.D.,1 Kent D. Bodily, Ph.D.,2 and Jeffrey S. Katz, Ph.D.3

Abstract

An interactive 3D desktop virtual environment task was created to investigate learning mechanisms in human problem solving. Participants were assessed for previous video game experience, divided into two groups (Training and Control), and matched for gender and experience. The Training group learned specific skills within the virtual environment before being presented a problem. The Control group was presented the problem only. Completion time was faster for the Training group and was affected by level of previous video game experience. Results indicated problem solving was a function of specific and general experience and demonstrated a method for dissociating these two facets of experience.

skills necessary to solve a novel problem and another group (Control group) did not. If the acquisition of specific skills facilitates problem solving, the Training group should complete the problem in less time than the Control group. The role of general experience was evaluated by assigning participants into matched groups based on an assessment of previous video game experience (PVE). This assessment occurred via questionnaire prior to participants’ participation in the task. If PVE facilitates problem solving in the virtual environment, a group consisting of participants with higher levels of PVE should complete the problem more quickly than a group of participants with lower levels of PVE. Additionally, the factors of specific and general experience may interact (i.e., differential influence of training dependent on level of general experience).

Introduction

D

ESPITE SUBTLE DEFINITIONAL NUANCES, the term problem solving generally refers to a process in which impediments are surmounted in order to reach a desired outcome.1–4 Therefore, problem solving serves many invaluable functions consisting of complex cognitive and behavioral components. Accordingly, an understanding of the processes and mechanisms that underlie problem solving is important in explaining these dynamically complex behaviors. Over the years, research has focused on three mechanisms integral to the problem-solving process: learning, representation, and choice.5 Research with humans focused on the learning mechanism has dwindled as research focused on representation and choice mechanisms has surged.6–17 Such oversight is perplexing considering the rich history of learning-centered research in the problem-solving realm. For example, many experiments conducted with nonhuman animals have demonstrated the influential role of learning in problem solving,18–23 and research has shown that experience is likely to serve the same critical roles for humans.24–28 In a novel attempt to account for prior learning experience with critical aspects of the test environment, the present study used a computer-generated 3D desktop virtual environment to investigate the learning mechanism in problem solving. The role of specific and general experience was tested by manipulating the acquisition of skills necessary for problem solving. Within the virtual environment, one group of participants (Training group) acquired specific component

Method Participants Thirty-two undergraduate students, 18 males and 14 females, enrolled in an introductory psychology courses served as participants. Each participant received extra credit for participation. Apparatus An interactive computer-generated 3D virtual environment was constructed and rendered using Valve Hammer Editor and run on the Half-Life Team Fortress Classic plat-

1Department

of Psychology, Armstrong Atlantic State University, Savannah, Georgia. of Psychology, Georgia Southern University, Statesboro, Georgia. 3Department of Psychology, Auburn University, Auburn, Alabama. 2Department

15

16

STURZ ET AL.

form. A personal computer with a KDS 17-in. flat-screen CRT monitor, Logitech optical mouse, Logitech quiet-touch keyboard, and SoundBlaster speakers served as the interface for the environment. The monitor (1152  864 pixels) provided participants with a first-person perspective1 of the virtual environment. The arrow keys of the keyboard, the mouse, and mouse buttons served to navigate and manipulate objects within the environment. Speakers were used to provide feedback to participants from the environment and to mask background noise. An identical second personal computer was utilized as the server for the virtual environment and recorded participants’ movements within the environment. All experimental events were controlled and recorded using Half-Life Television and Half-Life Dedicated Server on this second personal computer. Stimuli The stimuli were seven separate rooms. All rooms contained a floor, outer walls, and a ceiling and were well illuminated. All rooms also contained a green flashing light and an exit sign located just below the flashing light, which served as the location to which participants navigated. The Start Room (21.5  4.5  6 m)2 served as an information room that provided participants with instructions to navigate toward the green flashing light. In addition, it provided instructions that some tasks required key combinations. The Basic Navigation Room (navigational skills; 32.5  26  22 m) contained walls, raised above participants’ field of view, which formed a simple maze. The Pushing Boxes Room (pushing boxes skill; 32.5  26  22 m) contained three raised walls. The walls formed three barricades with an opening in each wall large enough through which participants could navigate. A box (1.6  1.6  1.6 m) was placed in front of each opening (total of three boxes). The Pulling Boxes Room (pulling boxes skill; 32.5  26  22 m) contained two walls, raised above the participants’ field of view, which partitioned the room into four distinct sections. The

FIG. 1.

walls formed three barricades with an opening in each wall large enough through which participants could navigate. In addition, two side walls formed a recessed area in front of each barricade. Inside each recessed area, one box was placed (total of three boxes). The Jumping Skill Room (jumping skill; 32.5  26  22 m) contained three gaps in the floor. Large street-type signs (0.8  1.6  1.6 m) served to warn participants of each of the gaps (total of six signs in this room). The Test Room (32.5  26  22 m) contained a large gap in the middle of the floor. Two large street-type signs warned participants of the gap. Four boxes were located in the two front corners of the room (total of eight boxes). A short wall ran alongside the boxes, parallel to the front wall, which reduced access to the boxes to only their sides. That is, boxes had to be pulled out of the recess created by the front wall of the room and the short wall (see Figure 1). The Goal Room (6.9  6.5  4 m) served as the room participants entered when the Test Room had been completed, and it contained messages informing participants that they had completed the task and instructing them to inform the investigator of their completion of the task. Design Each participant was screened using a questionnaire to measure PVE. Level of PVE was determined by multiplying the number of hours played per week by the number of years played. Participants were then sorted into categories of low (0), moderate (1–10), and high (10) experience. Balanced for experience, participants were randomly assigned to one of two groups: Training or Control. Each group contained 16 participants with an equal number of video game experienced and inexperienced participants. In addition, each group contained nine male and seven female participants. For males, each group contained three participants at the low, moderate, and high levels of PVE. For females, each group contained three participants at the low and moderate levels and one participant at the high level of PVE.3

Overview screenshot of the Test Room.

DISSOCIATION OF PAST AND PRESENT EXPERIENCE Procedure To minimize bias and reduce unnecessary exploratory behaviors, the informed consent notified all participants that the experiment was a test of spatial navigation and to proceed as quickly as possible. In the virtual environment, participants navigated toward a green flashing light and exit sign using the keyboard and mouse. Although participants were informed which keys and mouse buttons were active, they were not informed as to their specific functions. The arrow keys were used to move forward (↑), backward (↓), or side to side (← →). Auditory feedback was provided to indicate movement occurred (footstep sound). Moving the mouse left, right, forward, and backward rotated the view left, right, upward, and downward respectively. The left mouse button, if depressed and held, served to “hold onto” objects in the environment, and clicking the right mouse button served to jump. Auditory feedback (a huh sound) accompanied by the visual presentation of briefly rising above the ground indicated a jump had occurred. Before reaching the exit sign in the Start Room, instructions appeared on the screen informing participants to navigate toward the green flashing light in all rooms. In the Start Room, participants were also informed that some tasks required key combinations. After reaching the exit sign in the Start Room, participants were transported instantaneously to the next room. Participants were free to move about in the environment without any spatial or temporal constraints. However, once participants were transported into a room, they could not navigate to any previous room. Therefore, they were contained in the current room until they reached the exit sign. Training group. After reaching an exit sign, participants were transported instantaneously to the next room. The rooms were experienced in the following sequential order: Start Room, Basic Navigation Room, Pushing Boxes Room, Pulling Boxes Room, Jumping Skill Room, Test Room, and Goal Room. In the Basic Navigation Room, participants learned to navigate and operate controls. In the Pushing 18

Mean Completion Time in Test Room (in minutes)

16

17 Boxes Room, participants learned to push boxes out of the way and navigated through the space located in the wall. In the Pulling Boxes Room, participants learned to pull boxes out of recessed areas to reveal the spaces through which they must navigate. In the Jumping Skill Room, participants learned to jump over the three gaps. If participants fell into any of the gaps, they were immediately returned to the starting location of the same room. In the Test Room, the gap was too wide to jump to the other side. In order to complete this room, participants combined the skills acquired in the previous four rooms. Participants pulled boxes out of a recessed area, pushed the boxes into the gap, and jumped across on the boxes to the other side of the gap (see Figure 1). If participants fell into the gap, they were immediately returned to the starting location of this room. In the Goal Room, a message appeared on the screen informing participants to contact the investigator of their completion of the task. Control group. The procedure for the Control group was identical to the Training group with the exception that participants only experienced the Start Room, Test Room, and Goal Room. Thus, this group controlled for specific learning experiences in the virtual environment. Results Overall, the Training group (M  3.99 min, SEM  0.66) completed the Test Room faster than did the Control group (M  8.19 min, SEM  1.63), and males (M  3.46 min, SEM  0.62) completed the Test Room faster than did Females (M  9.48 min, SEM  1.63). In addition, mean completion time for the Test Room decreased as PVE increased: low (M  8.78 min, SEM  1.92), moderate (M  5.61 min, SEM  1.24), high (M  2.78 min, SEM  0.66). Despite males completing the Test Room faster than females, males and females exhibited a similar decrease in completion time for the Test Room as PVE increased. Figure 2 shows mean completion time for the Test Room for the three levels of PVE (low, moderate, high) by group. The effect of PVE was de-

Training Control

14 12 10 8 6 4 2 0 Low

Moderate

High

Previous Video Game Experience

FIG. 2. Mean completion time for the Test Room plotted across the three levels of previous video game experience (low, moderate, high) for each group (Training, Control). Error bars represent standard errors of the mean.

18 pendent upon group indicating that PVE did not influence performance of participants in the Training group but did in the Control group. A three-way ANOVA on completion time with group, gender, and PVE (low, moderate, high) as factors confirmed these results. There were main effects of group, F(1, 20)  9.81, p  0.01, 2  0.33; gender, F(1, 20)  19.46, p  0.001, 2  0.49; and PVE, F(2, 20)  6.27, p  0.01, 2  0.39. The group  PVE was the only significant interaction, F(2, 20)  5.04, p  0.05, 2  0.34, indicating that PVE was most influential on performance for the Control group. As a result of the interaction, custom contrasts were performed to identify differences in level of PVE within and across groups. Across groups, the low level showed a significant difference, F(1, 26)  12.35, p  0.05. No other significant across-group differences were found. Within the Control group, the low level was not significantly different from the moderate level, F(1, 13)  3.99, p  0.067, and was significantly different from the high level, F(1, 13)  8.99, p  0.05. The moderate level was not significantly different from the high level, F(1, 13)  1.47. Within the Training group, there were no significant differences. Discussion The present experiment found effects of both specific and general experience on problem solving. Participants trained to acquire specific component skills (i.e., navigational skills, pushing boxes, pulling boxes, jumping) were faster than participants who did not receive this training. Participants with higher levels of PVE (general experience) were faster than participants with lower levels of PVE. Overall, the present experiments provided evidence that problem solving is a function of both specific and general experience. Transfer of training and group differences in performance were obtained with the same procedure, suggesting these differences were a direct result of measuring and manipulating experience. Therefore, these findings are consistent with previous human and nonhuman animal studies that emphasized the influential role of experience in problem solving.18–28 Specifically, explicitly training component skills (i.e., experience resulting from specific interactions inside the experimental context, but not with the problem itself) facilitated solutions to a novel problem, and general experience (i.e., experience resulting from nonspecific interactions with stimuli outside of the experimental context) facilitated problem solving. Present results provided additional clarification by demonstrating a functional relationship between “general” experience and performance in a problem-solving task. Completion time systematically decreased with increasing levels of general experience. Most importantly, present results indicated an interaction between specific and general experience: the effect of training systematically decreased with increased levels of previous experience. Whereas previous studies have focused on the learning mechanism by exploring experience as a whole,18–28 the present study differentiated between specific skill acquisition and general past experience and highlighted their interaction. Therefore, the current methodology can be used to dissociate and control these two facets of experience and study their interaction. In conclusion, the contributions of the current experiments are twofold: methodological and empirical. The cur-

STURZ ET AL. rent approach to investigating the problem-solving process allowed precise control of experimental variables (e.g., measurement, identical arrangement and presentation of stimuli), control of experimental design (ensure differences are consistent across groups), and improved ecological validity (controlled problem-solving situation while maintaining realism). Our method permitted quantification of human prior experience in investigations of problem solving, and robust differences in problem-solving performance emerged. These strengths make it ripe for continued investigations of the learning mechanism and future investigations of the representational and choice mechanisms of problem solving. Notes 1. In first-person perspective, the monitor represents a view from the perspective of the character within the game; therefore, it represents a view of the virtual environment that is analogous to an individual’s view of the natural environment. 2. All dimensions are given as length by width by height and measured in meters (m). 3. Although a larger sample size for the female high level of PVE was preferred, great difficulties arose in obtaining females with high levels of such experience. Within the college population, there seemed to be fewer per capita than males. However, comparing the performance of males and females with high-level experience provided evidence to suggest that data of the female high level were not anomalistic and therefore warranted interpretation.

Disclosure Statement The authors have no conflict of interest. References 1. Duncker K. On problem solving. Psychological Monographs 1945; 58. 2. Halpern DF. (2003) Thought & knowledge: an introduction to critical thinking, 4th ed. Mahwah, NJ: Erlbaum. 3. Köhler W. (1927) The mentality of apes, 2nd ed. New York: Harcourt, Brace. 4. Newell A, Simon HA. (1972) Human problem solving. Englewood Cliffs, NJ: Prentice Hall. 5. Lovett MC. (2002) Problem solving. In Pashler H, Medin D, eds. Stevens’ handbook of experimental psychology: Memory & cognitive processes. New York: Wiley, pp. 317–62. 6. Best JB. The subgoal heuristic and its effect on internal representation. Journal of General Psychology 1987; 114:383–91. 7. Butler DL, Thomas KM. Preliminary study of the effectiveness of some heuristics used to generate solutions to ill-structured problems. Psychological Reports 1999; 84:817–27. 8. Genter D, Holyoak KJ. Reasoning and learning by analogy. American Psychologist 1997; 52:32–4. 9. Gigerenzer G, Hoffrage U. How to improve Bayesian reasoning without instruction. Psychological Review 1995; 102:684–704. 10. Grant ER, Spivey MJ. Eye movements and problem solving: guiding attention guides thought. Psychological Science 2003; 14:462–6. 11. Kotovsky K, Fallside D. (1989) Representation and transfer in problem solving. In Klahr D, Kotovsky K, eds. Complex information processing: the impact of Herbert A. Simon. Hillsdale, NJ: Erlbaum, pp. 69–108.

DISSOCIATION OF PAST AND PRESENT EXPERIENCE

19

12. Lovett MC, Schunn CD. Task representations, strategy variability, and base rate neglect. Journal of Experimental Psychology: General 1999; 128:107–30. 13. Metcalfe J. Feeling of knowing in memory and problem solving. Journal of Experimental Psychology: Learning, Memory, & Cognition 1986; 12:288–94. 14. Metcalfe J. Premonitions of insight predict impending error. Journal of Experimental Psychology: Learning, Memory, & Cognition 1986; 12:623–34. 15. Metcalfe J, Wiebe D. Intuition in insight and noninsight problem solving. Memory & Cognition 1987; 15:238–46. 16. Pretz JE, Naples JJ, Sternberg RJ. (2003) Recognizing, defining, and representing problems. In Davidson J, Sternberg R, eds. The psychology of problem solving. Cambridge: Cambridge University Press, pp 3–30. 17. Reynolds RI. The application of a search heuristic by skilled problem solvers. Bulletin of the Psychonomic Society 1991; 29:55–6. 18. Birch HG. The relation of previous experience to insightful problem-solving. Journal of Comparative Psychology 1945; 38:367–83. 19. Epstein R. The spontaneous interconnection of three repertoires. Psychological Record 1985; 35:131–41. 20. Epstein R, Kirshnit CE, Lanza RP, et al. “Insight” in the pigeon: antecedents and determinants of an intelligent performance. Nature 1984; 308:61–2. 21. Thorndike EL. Animal intelligence: an experimental study of the associative processes in animals. Psychological Review Monograph Supplement 1898; 2:1–109.

22. Thorndike EL. (1911) Animal intelligence: experimental studies. New York: Macmillan. 23. Thorndike EL. (1932) Fundamentals of learning. New York: Teachers College, Columbia University. 24. Jacobs MK, Dominowski RL. Learning to solve insight problems. Bulletin of the Psychonomic Society 1981; 17:171–4. 25. Lung C, Dominowski RR. Effects of strategy instruction on 9–dot problem solving. Journal of Experimental Psychology: Learning, Memory, & Cognition 1985; 11:804–11. 26. Weisberg RW, Alba JW. An examination of the alleged role of “fixation” in the solution of several “insight” problems. Journal of Experimental Psychology: General 1981; 110:169–92. 27. Weisberg RW, Alba JW. Gestalt theory, insight, and past experience: reply to Dominowski. Journal of Experimental Psychology: General 1981; 110:193–8. 28. Weisberg RW, Alba JW. Problem solving is not like perception: more on Gestalt theory. Journal of Experimental Psychology: General 1982; 111:326–30.

Address reprint requests to: Dr. Bradley R. Sturz Department of Psychology Armstrong Atlantic State University 229 Science Center 11935 Abercorn Street Savannah, GA 31419 E-mail: [email protected]