- Email: [email protected]
Artificial time constraints on the Iowa Gambling Task: The effects on behavioural performance and subjective experience
Brain and Cognition 57 (2005) 21–25 www.elsevier.com/locate/b&c
Artificial time constraints on the Iowa Gambling Task: The effects on behavioural performance and subjective experienceq Caroline H. Bowman, Cathryn E.Y. Evans, Oliver H. Turnbull* Centre for Cognitive Neuroscience, School of Psychology, University of Wales, Bangor, Wales LL57 2AS, United Kingdom Accepted 12 August 2004 Available online 5 November 2004
Abstract In the last decade, the Iowa Gambling Task (IGT) has become a widely employed neuropsychological research instrument for the investigation of executive function. The task has been employed in a wide range of formats, from ÔmanualÕ procedures to more recently introduced computerised versions. Computer-based formats often require that responses on the task should be artificially delayed by a number of seconds between trials to collect skin-conductance data. Participants, however, may become frustrated when they want to select from a particular deck in the time-limited versions—so that an unintended emotional experience of frustration might well disrupt a task presumed to be reliant on emotion-based learning. We investigated the effect of the various types of Iowa Gambling Task format on performance, using three types of task: the classic manual administration, with no time limitations; a computerised administration with a 6-s enforced delay; and a control computerised version which had no time constraints. We also evaluated the subjective experience of participants on each task. There were no significant differences in performance, between formats, in behavioural terms. Subjective experience measures on the task also showed consistent effects across all three formats—with substantial, and rapidly developing, awareness of which decks were ÔgoodÕ and Ôbad.Õ Ó 2004 Elsevier Inc. All rights reserved.
1. Introduction In the past decade the Iowa Gambling Task (IGT) has emerged as a widely employed neuropsychological research instrument in the investigation of executive function (e.g., Anderson, Bechara, Damasio, Tranel, & Damasio, 1999; Beninger et al., 2003; Cavallero et al., 2003; Crone, Vendel, & van der Molen, 2003; Suzuki, Hirota, Takasawa, & Shigemasu, in press). Since publication of the initial paper (Bechara, Damasio, Damasio, & Anderson, 1994), there have been almost 100 scientific studies that have employed the task, on a diverse set of neurological (e.g., Bechara, Damasio, Damasio, & Lee, q This research was funded by the National Alliance for Research on Schizophrenia and Depression (NARSAD) and the Wellcome Trust. * Corresponding author. Fax: +44 1248 382 599. E-mail address: [email protected] (O.H. Turnbull).
0278-2626/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bandc.2004.08.015
1999; Cavedini et al., 2002; North & OÕCarroll, 2001) and psychiatric populations (e.g., Blair, Colledge, & Mitchell, 2001; Schmitt, Brinkley, & Newman, 1999; Wilder, Weinberger, & Goldberg, 1998). The widespread use of the Iowa Gambling Task seems to be a result of the fact that it assesses an aspect of executive function that previously had been difficult to investigate (Anderson, Damasio, Damasio, & Tranel, 1988; Brickner, 1934; Damasio, Tranel, & Damasio, 1990; Eslinger & Damasio, 1985), but which is of considerable scientific interest (Bechara, 2003). It is usually regarded as an important measure of emotion-based learning (Damasio, 1994)—though see Tomb, Hauser, Deldin, and Caramazza (2002) for a contradictory view. The IGT requires the subject to repeatedly select one of four decks of cards, in any sequence, winning money with each card turn, but sometimes losing money on certain selections. Sustained playing of some decks leads to overall financial loss, while playing other decks leads to small but
22
C.H. Bowman et al. / Brain and Cognition 57 (2005) 21–25
consistent gains. Neurologically normal individuals gradually shift towards a preference for the Ôgood,Õ rather than the ÔbadÕ decks (Bechara et al., 1994). While there has been widespread use of the Iowa Gambling Task in the neuropsychological and neuropsychiatric literature, the task has been employed in a wide range of formats. One important dimension is the nature of reward on the task. The classic task (Bechara et al., 1994) involved the use of facsimile (or ÔmonopolyÕtype) money, which remains in use (Anderson, Bechara, Tranel, & Damasio, 1996; Bechara, Damasio, Tranel, & Anderson, 1998). However, in testing some populations, especially those with a psychiatric history, researchers have sometimes employed real money reinforcers (e.g., Blair et al., 2001; Bolla et al., 2003; Schmitt et al., 1999). No study had systematically investigated whether the nature of the reinforcer had an effect on performance on the task, until our group investigated the effect of real versus facsimile money (Bowman & Turnbull, 2003). We found no significant differences in performance, though the real money condition produced a narrower range of scores, and an associated smaller standard deviation (Bowman & Turnbull, 2003). On a task purporting to measure emotion-related effects, it is of considerable importance to ensure that variables that might influence the magnitude of emotion are well controlled. In this regard, it is of some interest to note that the nature of the reinforcer is not the only variable that has been altered in the many investigations of the task. While the standard procedure of unhindered choice amongst four decks of cards has remained consistent across almost all investigations (though see Bowman & Turnbull, in press), a key issue has been successive modifications of the technology employed in administering the Iowa Gambling Task. The initial investigations (Bechara et al., 1994, 1998; Bechara, Tranel, Damasio, & Damasio, 1996; Bechara, Damasio, Tranel, & Damasio, 1997) employed a ÔmanualÕ procedure, whereby the participant sat opposite the investigator, the decks were selected by physically turning cards, the investigator verbally described the outcome (‘‘win x dollars, but lose y dollars’’) and physical money (either real or facsimile) was moved across the desk. Later investigations (Bechara et al., 1999, Bechara, Dolan, & Hindes, 2002; Crone et al., 2003; Suzuki et al., in press) have employed a computerised version of the Iowa Gambling Task. Here the participant sits in front of a computer screen, chooses cards with a mouse-click, reads the outcome on the screen while hearing a tone that varies depending on whether the trial is a ÔwinÕ or a Ôloss,Õ and the financial aspect of the transaction is charted in terms of a virtual bar on the screen, which becomes longer or shorter as money is won or lost. In some respects, therefore, the manual versus computerised versions of the Iowa Gambling Task are quite different in their design. It is also of some note that no
papers have yet been published which investigate whether the two tasks differ in terms of the magnitude, or the rate, of learning. Might we anticipate that there would be performance differences across the different formats—given that many of the response properties, such as choice amongst four decks of cards, remain the same? In what way might the emotional, or reward, aspects of the task differ? In order to address this issue, we should recall that a central driving force behind the move from manual to computerised versions was the desire to collect skin-conductance data (Bechara et al., 1997, 1999). This requires accurate knowledge of the time course of events occurring on the task, so that a computerised system, which marks the occurrence of behavioural events on a skin-conductance trace, makes the analysis of such data more reliable. However, an additional factor associated with the collection of skinconductance data is the slow time course of measuring electrodermal responses (Bechara et al., 1999)—and this has shaped the administration of the Iowa Gambling Task in its computerised format. In the manual administration of the Iowa Gambling Task, participants are free to choose amongst decks across any time course. However, to collect accurate skin-conductance data, responses on the task need to be spaced, allowing a number of seconds to elapse between trials. In their first attempt to collect skin-conductance data, which they derived from a manual testing format, a minimum 15-s delay was introduced between card selections (Bechara et al., 1996, p. 221). When the Iowa group moved from manual to computerised administration (Bechara et al., 1999), they chose to enforce a time delay between card selections of six s. They made this decision based on previous reports that the average participant deliberation time prior to making a selection was roughly 10 s on the manual version of the task (Bechara et al., 2002, p. 1678). Why, then, might a restriction of decision time affect Iowa Gambling Task performance? In our initial administrations of a computerised version of the task, it became clear that administration typically took much longer in the computerised/time-delayed format. In the manual presentation, 100 card selections take the participant roughly 20–25 min, whereas in the computerised/time-delayed format, 100 selections take roughly 35–40 min. Also, it seemed that participants often became frustrated when they wanted to select a particular deck, but were not able to do so because of the enforced delay. Such feelings of frustration, i.e., a powerful emotion that was not intended to be part of test administration, might well disrupt a task presumed to be reliant on emotion-based learning (Bechara, 2003; Damasio, 1996). For these reasons, we considered it appropriate to investigate the effect of the various types of Iowa Gambling Task format on performance. We administered
C.H. Bowman et al. / Brain and Cognition 57 (2005) 21–25
three types of task. First, the classic manual administration (e.g., Bechara et al., 1994; Bowman & Turnbull, 2003; Wilder et al., 1998), with no time limitations on card choice. Second, the computerised administration, with the 6-s enforced delay (Bechara et al., 1999, 2002). These two versions of the task differ on two dimensions: namely manual versus computerised, and no time limitation versus time limitation. Therefore, we also evaluated a third type of task, a computerised version which had no time constraints. In addition to the standard behavioural measure of performance, we also evaluated the subjective experience of participants on each task (cf. Evans, Bowman, & Turnbull, under review).
2. Method 2.1. Participants Sixty-six undergraduates were recruited from the University of Wales, Bangor, each received course credit for participating. Additionally, participants were able to keep any money they won during the task. Twenty-two participants (all female, mean age 18.95 years) were administered the real money version of the manually presented IGT, as described previously (Bowman & Turnbull, 2003). Twenty-two participants (18 females, mean age 19.72 years; 4 males, mean age 19.5 years) were administered the computerised format of the IGT with no time constraints between card selections. Finally, 22 participants (16 females, mean age 19.13 years, 6 males, mean age 20.0 years) were administered a computerised administration of the IGT with a 6-s enforced delay between card selections. 2.2. Materials 2.2.1. Time-unlimited manual (TUM) Iowa Gambling Task The cards, money, and score sheet used in this study were those of the classic Iowa Gambling Task (Bechara et al., 1994), administered in the real money condition described elsewhere (Bowman & Turnbull, 2003). Participants were allowed to select cards freely throughout the duration of the 100 card selections. 2.2.2. Time-unlimited computerised (TUC) Iowa Gambling Task An entirely computerised version of the Iowa Gambling Task was programmed in the same manner as in the Bechara et al. (1999) study, and also administered in the real money condition (Bowman & Turnbull, 2003). The modified IGT Launcher, Version 1.0 (after Bechara et al., 1999) was installed on a Dell Inspiron 2650 laptop.
23
During this format of the IGT, the four decks of cards were represented virtually on the computer screen. In order to make a card selection, the participant clicked on their preferred choice using the computer mouse. After each selection was made, a visual display appeared on the screen indicating the amount of money won or lost as a result of that selection. An auditory tone, which signaled either a ÔwinÕ or a Ôloss,Õ accompanied the visual display. A green bar increased or decreased depending on money won or money lost, so that participants were aware of their overall standing. 2.2.3. Time-limited computerised (TLC) Iowa Gambling Task This version was identical to the time-unlimited computerised version, except that the computer posted a note requesting the participant to wait—an enforced delay that lasted for 6 s between card selections. During the time-delay period, all attempts at interacting with the computer, for example through mouse-click or button-press, were ineffective. 2.2.4. Subjective experience ratings After every block of card selections (where one block = 20 card selections), the game was interrupted briefly while participants were asked to provide subjective ratings regarding each deck of cards (decks A, B, C, and D), in terms of how ÔgoodÕ or ÔbadÕ they felt each deck was (where 0 = very bad and 10 = very good). In the manual task, the experimenter noted these responses. During the computerised presentation of the task, the questions and feedback were presented on screen, and participants typed their responses to each of the questions. Due to technical difficulties, the subjective experience data from four participants in the TUC task, and one participant in the TLC group, could not be analysed.
3. Results As in Bechara et al. (1994), the 100 card selections from the original gambling task were sub-divided into five blocks: 1–20, 21–40, 41–60, 61–80, and 81–100. The net score for each block was calculated by subtracting the number of bad selections from the number of good selections: [(C + D) (A + B)]. In each stage, a net score below zero represented that participants were selecting disadvantageously, and a score above zero reflected advantageous choices. Additionally, the subjective experience ratings were analysed in the same manner, that is, for each block a net score was calculated by subtracting total number of bad deck ratings from total number of good deck ratings. Fig. 1 illustrates the performance of the three groups of participants (N = 22 in each group). Regardless of
24
C.H. Bowman et al. / Brain and Cognition 57 (2005) 21–25
In terms of subjective experience (see Fig. 2) there was a rapid increase in ratings of the ÔgoodÕ decks between blocks one and two. From block two onwards, the ratings remained relatively stable over the remaining selections. It is notable that the subjective experience ratings of the good decks began, and remained, higher in the TUC group than in either the TLC or the TUM group. A mixed ANOVA revealed no main effect of Group (F (2, 58) = 2.84, p > .05). There was a main effect of Block (F (4, 232) = 13.76, p < .001), but no interaction between Group and Block (F (8, 232) = .20, p > .05).
4. Discussion
Fig. 1. Mean number of good–bad card selections per block for each group of participants.
group, participants began with a performance level below chance, and thereafter were able to choose increasingly more cards from the ÔgoodÕ decks as opposed to the ÔbadÕ decks. A mixed ANOVA did not find a main effect of Group (F (2, 63) = .324, p > .05). There was a main effect of Block, (F (4, 252) = 20.91, p < .001) as expected, but no interaction between Group and Block (F (8, 25) = .55, p > .05).
Fig. 2. Mean number of good–bad (g–b) deck subjective experience ratings per block for each group of participants.
A profile of rapid learning on the Iowa Gambling Task was found, for all administration formats, starting with performance levels below chance, and improving to a level of +2 to +5. This finding is consistent with a series of papers on the Iowa Gambling Task, in which similar patterns of learning are found across a range of reward profiles (e.g., Bowman & Turnbull, 2003; Schmitt et al., 1999) and administration techniques (Bechara et al., 1994, 1999, 2002). It is of some note that the final levels of behavioural performance achieved (+2 to +5) were consistent with the previous findings of our group (Evans, Kemish, & Turnbull, in press) in which participants with university education show learning to roughly this level. A key finding of the present study is that there were no significant differences in performance, between formats, in behavioural terms. It appears that the possible effects of frustration, in the time-limited format, have no bearing on behavioural performance. Thus, the principal finding of the present study is that the format of administration (manual versus computerised, and timelimited versus time-unlimited), does not appear to affect IGT performance. Subjective experience measures on the task also show consistent effects across all three formats. Participants showed substantial, and rapidly developing, awareness of which decks were ÔgoodÕ and ÔbadÕ—with subjective experience ratings at above-chance levels in some instances even after the first 20 trials, and invariably well above chance levels after 40 trials. These data are consistent with the findings of Evans et al. (under review). Together, these data suggest that participants had greater awareness of the nature of the task than their behavioural performance demonstrates, a claim that would run counter to that proposed by Bechara et al. (1997, 2000), who suggested that the IGT can be successfully performed without fully developed awareness of the way in which the game operates, and the emotional valence associated with individual decks (see Evans et al., under review, for more discussion on this point).
C.H. Bowman et al. / Brain and Cognition 57 (2005) 21–25
Thus, it appears that the enforcement of an artificial time-delay into the Iowa Gambling Task may not produce a decrement in learning, as measured by the basic behavioural performance. This is consistent with findings, reported elsewhere, that additional mental operations, such as completing a cognitively demanding secondary task, have little effect on behavioural performance on the IGT (Turnbull, Evans, Bunce, Carzolio, & OÕConnor, in press). References Anderson, S. W., Bechara, A., Damasio, H., Tranel, D., & Damasio, A. R. (1999). Impairment of social and moral behaviour related to early damage in human prefrontal cortex. Nature Neuroscience, 2, 1032–1037. Anderson, S. W., Bechara, A., Tranel, D., & Damasio, H. (1996). Characterization of the decision-making defect of subjects with ventromedial frontal lobe damage. Society for Neuroscience, 22, 1108. Anderson, S. W., Damasio, H., Damasio, A. R., & Tranel, D. (1988). Neuropsychological correlates of bilateral frontal lobe lesions in humans. Society for Neuroscience, 14, 1288. Bechara, A. (2003). Risky business: Emotion, decision-making, and addiction. Journal of Gambling Studies, 19, 23–51. Bechara, A., Damasio, A. R., Damasio, H., & Anderson, S. W. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition, 50, 7–15. Bechara, A., Damasio, H., Damasio, A. R., & Lee, G. P. (1999). Different contributions of the human amygdala and ventromedial prefrontal cortex to decision-making. Journal of Neuroscience, 19, 5473–5481. Bechara, A., Damasio, H., Tranel, D., & Anderson, S. W. (1998). Dissociation of working memory from decision-making within the human prefrontal cortex. Journal of Neuroscience, 18, 428–437. Bechara, A., Damasio, H., Tranel, D., & Damasio, A. R. (1997). Deciding advantageously before knowing the advantageous strategy. Science, 275, 1293–1294. Bechara, A., Dolan, S., & Hindes, A. (2002). Decision-making and addiction (part II): Myopia for the future or hypersensitivity to reward?. Neuropsychologia, 40, 1690–1705. Bechara, A., Tranel, D., Damasio, H., & Damasio, A. R. (1996). Failure to respond autonomically to anticipated future outcomes following damage to prefrontal cortex. Cerebral Cortex, 6(2), 215–225. Beninger, R. J., Wasserman, J., Zanibbi, K., Charbonneau, D., Mangels, J., & Beninger, B. V. (2003). Typical and atypical antipsychotic medications differentially affect two nondeclarative memory tasks in schizophrenic patients: A double dissociation. Schizophrenia Research, 61, 281–292. Blair, R. J. R, Colledge, E., & Mitchell, D. G. V. (2001). Somatic markers and response reversal: Is there orbitofrontal cortex dysfunction in boys with psychopathic tendencies?. Journal of Abnormal Child Psychology, 29, 499–511. Bolla, K. I., Eldreth, D. A., London, E. D., Kiehl, K. A., Mouratidis, M., Contoreggi, C., et al. (2003). Orbitofrontal cortex dysfunction in abstinent cocaine abusers performing a decision-making task. NeuroImage, 19, 1085–1094.
25
Bowman, C. H., & Turnbull, O. H. (in press). Emotion-based learning on a simplified card task: The Iowa and Bangor Gambling Tasks. Brain and Cognition. Bowman, C. H., & Turnbull, O. H. (2003). Real versus facsimile reinforcers on the Iowa Gambling Task. Brain and Cognition, 53, 207–210. Brickner, R. M. (1934). An interpretation of frontal lobe function based upon the study of a case of partial bilateral frontal lobectomy. In S. Anand (Ed.), Research publications. New York: Association for Research in Nervous and Mental Diseases. Cavallero, R., Cavedini, P., Mistretta, P., Bassi, T., Angelone, S. M., Ubbiali, A., et al. (2003). Basal-corticofrontal circuits in schizophrenia and obsessive–compulsive disorder: A controlled doubledissociation study. Biological Psychiatry, 54, 437–443. Cavedini, P., Riboldi, G., DÕAnnucci, A., Belotti, P., Cisima, M., & Bellodi, L. (2002). Decision-making heterogeneity in obsessive– compulsive disorder: Ventromedial prefrontal cortex function predicts different treatment outcomes. Neuropsychologia, 40, 205–211. Crone, E. A., Vendel, I., & van der Molen, M. W. (2003). Decisionmaking in disinhibited adolescents and adults: Insensitivity to future consequences or driven by immediate reward?. Personality and Individual Differences, 34, 1–17. Damasio, A. R. (1994). Descartes Error: Emotion, reason, and the human brain. New York: Grosset/Putnam. Damasio, A. R. (1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex. Philosophical Transactions of the Royal Society of London B, 351, 1413–1420. Damasio, A. R., Tranel, D., & Damasio, H. (1990). Individuals with sociopathic behaviour caused by frontal damage fail to respond autonomically to social stimuli. Behavioural Brain Research, 41, 81–94. Eslinger, P. J., & Damasio, A. R. (1985). Severe disturbance of higher cognition after bilateral frontal lobe ablation: Patient EVR. Neurology, 35, 1731–1741. Evans, C. E. Y., Bowman, C. H., & Turnbull, O. H. (under review). Subjective experience and the Iowa Gambling Task: Preserved performance in schizophrenia. Evans, C. E. Y., Kemish, K., & Turnbull, O. H. (in press). Paradoxical effects of education on the Iowa Gambling Task. Brain and Cognition. North, N. T., & OÕCarroll, R. E. (2001). Decision making in patients with spinal cord damage: Afferent feedback and the somatic marker hypothesis. Neuropsychologia, 39, 521–524. Schmitt, W. A., Brinkley, C. A., & Newman, J. P. (1999). Testing DamasioÕs somatic marker hypothesis with psychopathic individuals: Risk takers or risk averse?. Journal of Abnormal Psychology, 3, 538–543. Suzuki, A., Hirota, A., Takasawa, N., & Shigemasu, K. (in press). Application of the somatic marker hypothesis to individual differences in decision making. Biological Psychiatry. Tomb, I., Hauser, M., Deldin, P., & Caramazza, A. (2002). Do somatic markers mediate decisions on the gambling task?. Nature Neuroscience, 5, 1103–1104. Turnbull, O. H., Evans, C. E. Y., Bunce, A., Carzolio, B., & OÕConnor, J. (in press). Central executive resources and the Iowa Gambling Task. Brain and Cognition. Wilder, K. E., Weinberger, D. R., & Goldberg, T. E. (1998). Operant conditioning and the orbitofrontal cortex in schizophrenic patients: Unexpected evidence for intact functioning. Schizophrenia Research, 30, 169–174.
Artificial time constraints on the Iowa Gambling Task: The effects on behavioural performance and subjective experienceq Caroline H. Bowman, Cathryn E.Y. Evans, Oliver H. Turnbull* Centre for Cognitive Neuroscience, School of Psychology, University of Wales, Bangor, Wales LL57 2AS, United Kingdom Accepted 12 August 2004 Available online 5 November 2004
Abstract In the last decade, the Iowa Gambling Task (IGT) has become a widely employed neuropsychological research instrument for the investigation of executive function. The task has been employed in a wide range of formats, from ÔmanualÕ procedures to more recently introduced computerised versions. Computer-based formats often require that responses on the task should be artificially delayed by a number of seconds between trials to collect skin-conductance data. Participants, however, may become frustrated when they want to select from a particular deck in the time-limited versions—so that an unintended emotional experience of frustration might well disrupt a task presumed to be reliant on emotion-based learning. We investigated the effect of the various types of Iowa Gambling Task format on performance, using three types of task: the classic manual administration, with no time limitations; a computerised administration with a 6-s enforced delay; and a control computerised version which had no time constraints. We also evaluated the subjective experience of participants on each task. There were no significant differences in performance, between formats, in behavioural terms. Subjective experience measures on the task also showed consistent effects across all three formats—with substantial, and rapidly developing, awareness of which decks were ÔgoodÕ and Ôbad.Õ Ó 2004 Elsevier Inc. All rights reserved.
1. Introduction In the past decade the Iowa Gambling Task (IGT) has emerged as a widely employed neuropsychological research instrument in the investigation of executive function (e.g., Anderson, Bechara, Damasio, Tranel, & Damasio, 1999; Beninger et al., 2003; Cavallero et al., 2003; Crone, Vendel, & van der Molen, 2003; Suzuki, Hirota, Takasawa, & Shigemasu, in press). Since publication of the initial paper (Bechara, Damasio, Damasio, & Anderson, 1994), there have been almost 100 scientific studies that have employed the task, on a diverse set of neurological (e.g., Bechara, Damasio, Damasio, & Lee, q This research was funded by the National Alliance for Research on Schizophrenia and Depression (NARSAD) and the Wellcome Trust. * Corresponding author. Fax: +44 1248 382 599. E-mail address: [email protected] (O.H. Turnbull).
0278-2626/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bandc.2004.08.015
1999; Cavedini et al., 2002; North & OÕCarroll, 2001) and psychiatric populations (e.g., Blair, Colledge, & Mitchell, 2001; Schmitt, Brinkley, & Newman, 1999; Wilder, Weinberger, & Goldberg, 1998). The widespread use of the Iowa Gambling Task seems to be a result of the fact that it assesses an aspect of executive function that previously had been difficult to investigate (Anderson, Damasio, Damasio, & Tranel, 1988; Brickner, 1934; Damasio, Tranel, & Damasio, 1990; Eslinger & Damasio, 1985), but which is of considerable scientific interest (Bechara, 2003). It is usually regarded as an important measure of emotion-based learning (Damasio, 1994)—though see Tomb, Hauser, Deldin, and Caramazza (2002) for a contradictory view. The IGT requires the subject to repeatedly select one of four decks of cards, in any sequence, winning money with each card turn, but sometimes losing money on certain selections. Sustained playing of some decks leads to overall financial loss, while playing other decks leads to small but
22
C.H. Bowman et al. / Brain and Cognition 57 (2005) 21–25
consistent gains. Neurologically normal individuals gradually shift towards a preference for the Ôgood,Õ rather than the ÔbadÕ decks (Bechara et al., 1994). While there has been widespread use of the Iowa Gambling Task in the neuropsychological and neuropsychiatric literature, the task has been employed in a wide range of formats. One important dimension is the nature of reward on the task. The classic task (Bechara et al., 1994) involved the use of facsimile (or ÔmonopolyÕtype) money, which remains in use (Anderson, Bechara, Tranel, & Damasio, 1996; Bechara, Damasio, Tranel, & Anderson, 1998). However, in testing some populations, especially those with a psychiatric history, researchers have sometimes employed real money reinforcers (e.g., Blair et al., 2001; Bolla et al., 2003; Schmitt et al., 1999). No study had systematically investigated whether the nature of the reinforcer had an effect on performance on the task, until our group investigated the effect of real versus facsimile money (Bowman & Turnbull, 2003). We found no significant differences in performance, though the real money condition produced a narrower range of scores, and an associated smaller standard deviation (Bowman & Turnbull, 2003). On a task purporting to measure emotion-related effects, it is of considerable importance to ensure that variables that might influence the magnitude of emotion are well controlled. In this regard, it is of some interest to note that the nature of the reinforcer is not the only variable that has been altered in the many investigations of the task. While the standard procedure of unhindered choice amongst four decks of cards has remained consistent across almost all investigations (though see Bowman & Turnbull, in press), a key issue has been successive modifications of the technology employed in administering the Iowa Gambling Task. The initial investigations (Bechara et al., 1994, 1998; Bechara, Tranel, Damasio, & Damasio, 1996; Bechara, Damasio, Tranel, & Damasio, 1997) employed a ÔmanualÕ procedure, whereby the participant sat opposite the investigator, the decks were selected by physically turning cards, the investigator verbally described the outcome (‘‘win x dollars, but lose y dollars’’) and physical money (either real or facsimile) was moved across the desk. Later investigations (Bechara et al., 1999, Bechara, Dolan, & Hindes, 2002; Crone et al., 2003; Suzuki et al., in press) have employed a computerised version of the Iowa Gambling Task. Here the participant sits in front of a computer screen, chooses cards with a mouse-click, reads the outcome on the screen while hearing a tone that varies depending on whether the trial is a ÔwinÕ or a Ôloss,Õ and the financial aspect of the transaction is charted in terms of a virtual bar on the screen, which becomes longer or shorter as money is won or lost. In some respects, therefore, the manual versus computerised versions of the Iowa Gambling Task are quite different in their design. It is also of some note that no
papers have yet been published which investigate whether the two tasks differ in terms of the magnitude, or the rate, of learning. Might we anticipate that there would be performance differences across the different formats—given that many of the response properties, such as choice amongst four decks of cards, remain the same? In what way might the emotional, or reward, aspects of the task differ? In order to address this issue, we should recall that a central driving force behind the move from manual to computerised versions was the desire to collect skin-conductance data (Bechara et al., 1997, 1999). This requires accurate knowledge of the time course of events occurring on the task, so that a computerised system, which marks the occurrence of behavioural events on a skin-conductance trace, makes the analysis of such data more reliable. However, an additional factor associated with the collection of skinconductance data is the slow time course of measuring electrodermal responses (Bechara et al., 1999)—and this has shaped the administration of the Iowa Gambling Task in its computerised format. In the manual administration of the Iowa Gambling Task, participants are free to choose amongst decks across any time course. However, to collect accurate skin-conductance data, responses on the task need to be spaced, allowing a number of seconds to elapse between trials. In their first attempt to collect skin-conductance data, which they derived from a manual testing format, a minimum 15-s delay was introduced between card selections (Bechara et al., 1996, p. 221). When the Iowa group moved from manual to computerised administration (Bechara et al., 1999), they chose to enforce a time delay between card selections of six s. They made this decision based on previous reports that the average participant deliberation time prior to making a selection was roughly 10 s on the manual version of the task (Bechara et al., 2002, p. 1678). Why, then, might a restriction of decision time affect Iowa Gambling Task performance? In our initial administrations of a computerised version of the task, it became clear that administration typically took much longer in the computerised/time-delayed format. In the manual presentation, 100 card selections take the participant roughly 20–25 min, whereas in the computerised/time-delayed format, 100 selections take roughly 35–40 min. Also, it seemed that participants often became frustrated when they wanted to select a particular deck, but were not able to do so because of the enforced delay. Such feelings of frustration, i.e., a powerful emotion that was not intended to be part of test administration, might well disrupt a task presumed to be reliant on emotion-based learning (Bechara, 2003; Damasio, 1996). For these reasons, we considered it appropriate to investigate the effect of the various types of Iowa Gambling Task format on performance. We administered
C.H. Bowman et al. / Brain and Cognition 57 (2005) 21–25
three types of task. First, the classic manual administration (e.g., Bechara et al., 1994; Bowman & Turnbull, 2003; Wilder et al., 1998), with no time limitations on card choice. Second, the computerised administration, with the 6-s enforced delay (Bechara et al., 1999, 2002). These two versions of the task differ on two dimensions: namely manual versus computerised, and no time limitation versus time limitation. Therefore, we also evaluated a third type of task, a computerised version which had no time constraints. In addition to the standard behavioural measure of performance, we also evaluated the subjective experience of participants on each task (cf. Evans, Bowman, & Turnbull, under review).
2. Method 2.1. Participants Sixty-six undergraduates were recruited from the University of Wales, Bangor, each received course credit for participating. Additionally, participants were able to keep any money they won during the task. Twenty-two participants (all female, mean age 18.95 years) were administered the real money version of the manually presented IGT, as described previously (Bowman & Turnbull, 2003). Twenty-two participants (18 females, mean age 19.72 years; 4 males, mean age 19.5 years) were administered the computerised format of the IGT with no time constraints between card selections. Finally, 22 participants (16 females, mean age 19.13 years, 6 males, mean age 20.0 years) were administered a computerised administration of the IGT with a 6-s enforced delay between card selections. 2.2. Materials 2.2.1. Time-unlimited manual (TUM) Iowa Gambling Task The cards, money, and score sheet used in this study were those of the classic Iowa Gambling Task (Bechara et al., 1994), administered in the real money condition described elsewhere (Bowman & Turnbull, 2003). Participants were allowed to select cards freely throughout the duration of the 100 card selections. 2.2.2. Time-unlimited computerised (TUC) Iowa Gambling Task An entirely computerised version of the Iowa Gambling Task was programmed in the same manner as in the Bechara et al. (1999) study, and also administered in the real money condition (Bowman & Turnbull, 2003). The modified IGT Launcher, Version 1.0 (after Bechara et al., 1999) was installed on a Dell Inspiron 2650 laptop.
23
During this format of the IGT, the four decks of cards were represented virtually on the computer screen. In order to make a card selection, the participant clicked on their preferred choice using the computer mouse. After each selection was made, a visual display appeared on the screen indicating the amount of money won or lost as a result of that selection. An auditory tone, which signaled either a ÔwinÕ or a Ôloss,Õ accompanied the visual display. A green bar increased or decreased depending on money won or money lost, so that participants were aware of their overall standing. 2.2.3. Time-limited computerised (TLC) Iowa Gambling Task This version was identical to the time-unlimited computerised version, except that the computer posted a note requesting the participant to wait—an enforced delay that lasted for 6 s between card selections. During the time-delay period, all attempts at interacting with the computer, for example through mouse-click or button-press, were ineffective. 2.2.4. Subjective experience ratings After every block of card selections (where one block = 20 card selections), the game was interrupted briefly while participants were asked to provide subjective ratings regarding each deck of cards (decks A, B, C, and D), in terms of how ÔgoodÕ or ÔbadÕ they felt each deck was (where 0 = very bad and 10 = very good). In the manual task, the experimenter noted these responses. During the computerised presentation of the task, the questions and feedback were presented on screen, and participants typed their responses to each of the questions. Due to technical difficulties, the subjective experience data from four participants in the TUC task, and one participant in the TLC group, could not be analysed.
3. Results As in Bechara et al. (1994), the 100 card selections from the original gambling task were sub-divided into five blocks: 1–20, 21–40, 41–60, 61–80, and 81–100. The net score for each block was calculated by subtracting the number of bad selections from the number of good selections: [(C + D) (A + B)]. In each stage, a net score below zero represented that participants were selecting disadvantageously, and a score above zero reflected advantageous choices. Additionally, the subjective experience ratings were analysed in the same manner, that is, for each block a net score was calculated by subtracting total number of bad deck ratings from total number of good deck ratings. Fig. 1 illustrates the performance of the three groups of participants (N = 22 in each group). Regardless of
24
C.H. Bowman et al. / Brain and Cognition 57 (2005) 21–25
In terms of subjective experience (see Fig. 2) there was a rapid increase in ratings of the ÔgoodÕ decks between blocks one and two. From block two onwards, the ratings remained relatively stable over the remaining selections. It is notable that the subjective experience ratings of the good decks began, and remained, higher in the TUC group than in either the TLC or the TUM group. A mixed ANOVA revealed no main effect of Group (F (2, 58) = 2.84, p > .05). There was a main effect of Block (F (4, 232) = 13.76, p < .001), but no interaction between Group and Block (F (8, 232) = .20, p > .05).
4. Discussion
Fig. 1. Mean number of good–bad card selections per block for each group of participants.
group, participants began with a performance level below chance, and thereafter were able to choose increasingly more cards from the ÔgoodÕ decks as opposed to the ÔbadÕ decks. A mixed ANOVA did not find a main effect of Group (F (2, 63) = .324, p > .05). There was a main effect of Block, (F (4, 252) = 20.91, p < .001) as expected, but no interaction between Group and Block (F (8, 25) = .55, p > .05).
Fig. 2. Mean number of good–bad (g–b) deck subjective experience ratings per block for each group of participants.
A profile of rapid learning on the Iowa Gambling Task was found, for all administration formats, starting with performance levels below chance, and improving to a level of +2 to +5. This finding is consistent with a series of papers on the Iowa Gambling Task, in which similar patterns of learning are found across a range of reward profiles (e.g., Bowman & Turnbull, 2003; Schmitt et al., 1999) and administration techniques (Bechara et al., 1994, 1999, 2002). It is of some note that the final levels of behavioural performance achieved (+2 to +5) were consistent with the previous findings of our group (Evans, Kemish, & Turnbull, in press) in which participants with university education show learning to roughly this level. A key finding of the present study is that there were no significant differences in performance, between formats, in behavioural terms. It appears that the possible effects of frustration, in the time-limited format, have no bearing on behavioural performance. Thus, the principal finding of the present study is that the format of administration (manual versus computerised, and timelimited versus time-unlimited), does not appear to affect IGT performance. Subjective experience measures on the task also show consistent effects across all three formats. Participants showed substantial, and rapidly developing, awareness of which decks were ÔgoodÕ and ÔbadÕ—with subjective experience ratings at above-chance levels in some instances even after the first 20 trials, and invariably well above chance levels after 40 trials. These data are consistent with the findings of Evans et al. (under review). Together, these data suggest that participants had greater awareness of the nature of the task than their behavioural performance demonstrates, a claim that would run counter to that proposed by Bechara et al. (1997, 2000), who suggested that the IGT can be successfully performed without fully developed awareness of the way in which the game operates, and the emotional valence associated with individual decks (see Evans et al., under review, for more discussion on this point).
C.H. Bowman et al. / Brain and Cognition 57 (2005) 21–25
Thus, it appears that the enforcement of an artificial time-delay into the Iowa Gambling Task may not produce a decrement in learning, as measured by the basic behavioural performance. This is consistent with findings, reported elsewhere, that additional mental operations, such as completing a cognitively demanding secondary task, have little effect on behavioural performance on the IGT (Turnbull, Evans, Bunce, Carzolio, & OÕConnor, in press). References Anderson, S. W., Bechara, A., Damasio, H., Tranel, D., & Damasio, A. R. (1999). Impairment of social and moral behaviour related to early damage in human prefrontal cortex. Nature Neuroscience, 2, 1032–1037. Anderson, S. W., Bechara, A., Tranel, D., & Damasio, H. (1996). Characterization of the decision-making defect of subjects with ventromedial frontal lobe damage. Society for Neuroscience, 22, 1108. Anderson, S. W., Damasio, H., Damasio, A. R., & Tranel, D. (1988). Neuropsychological correlates of bilateral frontal lobe lesions in humans. Society for Neuroscience, 14, 1288. Bechara, A. (2003). Risky business: Emotion, decision-making, and addiction. Journal of Gambling Studies, 19, 23–51. Bechara, A., Damasio, A. R., Damasio, H., & Anderson, S. W. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition, 50, 7–15. Bechara, A., Damasio, H., Damasio, A. R., & Lee, G. P. (1999). Different contributions of the human amygdala and ventromedial prefrontal cortex to decision-making. Journal of Neuroscience, 19, 5473–5481. Bechara, A., Damasio, H., Tranel, D., & Anderson, S. W. (1998). Dissociation of working memory from decision-making within the human prefrontal cortex. Journal of Neuroscience, 18, 428–437. Bechara, A., Damasio, H., Tranel, D., & Damasio, A. R. (1997). Deciding advantageously before knowing the advantageous strategy. Science, 275, 1293–1294. Bechara, A., Dolan, S., & Hindes, A. (2002). Decision-making and addiction (part II): Myopia for the future or hypersensitivity to reward?. Neuropsychologia, 40, 1690–1705. Bechara, A., Tranel, D., Damasio, H., & Damasio, A. R. (1996). Failure to respond autonomically to anticipated future outcomes following damage to prefrontal cortex. Cerebral Cortex, 6(2), 215–225. Beninger, R. J., Wasserman, J., Zanibbi, K., Charbonneau, D., Mangels, J., & Beninger, B. V. (2003). Typical and atypical antipsychotic medications differentially affect two nondeclarative memory tasks in schizophrenic patients: A double dissociation. Schizophrenia Research, 61, 281–292. Blair, R. J. R, Colledge, E., & Mitchell, D. G. V. (2001). Somatic markers and response reversal: Is there orbitofrontal cortex dysfunction in boys with psychopathic tendencies?. Journal of Abnormal Child Psychology, 29, 499–511. Bolla, K. I., Eldreth, D. A., London, E. D., Kiehl, K. A., Mouratidis, M., Contoreggi, C., et al. (2003). Orbitofrontal cortex dysfunction in abstinent cocaine abusers performing a decision-making task. NeuroImage, 19, 1085–1094.
25
Bowman, C. H., & Turnbull, O. H. (in press). Emotion-based learning on a simplified card task: The Iowa and Bangor Gambling Tasks. Brain and Cognition. Bowman, C. H., & Turnbull, O. H. (2003). Real versus facsimile reinforcers on the Iowa Gambling Task. Brain and Cognition, 53, 207–210. Brickner, R. M. (1934). An interpretation of frontal lobe function based upon the study of a case of partial bilateral frontal lobectomy. In S. Anand (Ed.), Research publications. New York: Association for Research in Nervous and Mental Diseases. Cavallero, R., Cavedini, P., Mistretta, P., Bassi, T., Angelone, S. M., Ubbiali, A., et al. (2003). Basal-corticofrontal circuits in schizophrenia and obsessive–compulsive disorder: A controlled doubledissociation study. Biological Psychiatry, 54, 437–443. Cavedini, P., Riboldi, G., DÕAnnucci, A., Belotti, P., Cisima, M., & Bellodi, L. (2002). Decision-making heterogeneity in obsessive– compulsive disorder: Ventromedial prefrontal cortex function predicts different treatment outcomes. Neuropsychologia, 40, 205–211. Crone, E. A., Vendel, I., & van der Molen, M. W. (2003). Decisionmaking in disinhibited adolescents and adults: Insensitivity to future consequences or driven by immediate reward?. Personality and Individual Differences, 34, 1–17. Damasio, A. R. (1994). Descartes Error: Emotion, reason, and the human brain. New York: Grosset/Putnam. Damasio, A. R. (1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex. Philosophical Transactions of the Royal Society of London B, 351, 1413–1420. Damasio, A. R., Tranel, D., & Damasio, H. (1990). Individuals with sociopathic behaviour caused by frontal damage fail to respond autonomically to social stimuli. Behavioural Brain Research, 41, 81–94. Eslinger, P. J., & Damasio, A. R. (1985). Severe disturbance of higher cognition after bilateral frontal lobe ablation: Patient EVR. Neurology, 35, 1731–1741. Evans, C. E. Y., Bowman, C. H., & Turnbull, O. H. (under review). Subjective experience and the Iowa Gambling Task: Preserved performance in schizophrenia. Evans, C. E. Y., Kemish, K., & Turnbull, O. H. (in press). Paradoxical effects of education on the Iowa Gambling Task. Brain and Cognition. North, N. T., & OÕCarroll, R. E. (2001). Decision making in patients with spinal cord damage: Afferent feedback and the somatic marker hypothesis. Neuropsychologia, 39, 521–524. Schmitt, W. A., Brinkley, C. A., & Newman, J. P. (1999). Testing DamasioÕs somatic marker hypothesis with psychopathic individuals: Risk takers or risk averse?. Journal of Abnormal Psychology, 3, 538–543. Suzuki, A., Hirota, A., Takasawa, N., & Shigemasu, K. (in press). Application of the somatic marker hypothesis to individual differences in decision making. Biological Psychiatry. Tomb, I., Hauser, M., Deldin, P., & Caramazza, A. (2002). Do somatic markers mediate decisions on the gambling task?. Nature Neuroscience, 5, 1103–1104. Turnbull, O. H., Evans, C. E. Y., Bunce, A., Carzolio, B., & OÕConnor, J. (in press). Central executive resources and the Iowa Gambling Task. Brain and Cognition. Wilder, K. E., Weinberger, D. R., & Goldberg, T. E. (1998). Operant conditioning and the orbitofrontal cortex in schizophrenic patients: Unexpected evidence for intact functioning. Schizophrenia Research, 30, 169–174.