PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of acnLink to Publisher's site
 
Arch Clin Neuropsychol. Feb 2010; 25(1): 28–38.
Published online Nov 25, 2009. doi:  10.1093/arclin/acp094
PMCID: PMC2809553
Influence of Procedural Learning on Iowa Gambling Task Performance Among HIV+ Individuals with History of Substance Dependence
Raul Gonzalez,a* Margaret Wardle,b Joanna Jacobus,cd Jasmin Vassileva,a and Eileen M. Martin-Thormeyerae
aDepartment of Psychiatry, University of Illinois-Chicago, Chicago, IL, USA
bDepartment of Psychiatry, University of Chicago, Chicago, IL, USA
cDepartment of Psychology, San Diego State University, San Diego, CA, USA
dDepartment of Psychiatry, University of California, San Diego Joint Doctoral Program in Clinical Psychology, San Diego, CA, USA
eJesse Brown VA Medical Center, Chicago, IL, USA
*Corresponding author at: Department of Psychiatry, University of Illinois-Chicago, 1601 W. Taylor St MC 912, Chicago, IL 60622, USA. Tel.: Phone: +1-312-413-5956; fax: +1-312-413-8147. E-mail address:rgonzalez/at/psych.uic.edu (R. Gonzalez).
Accepted October 29, 2009.
HIV+ individuals have been shown to demonstrate deficits on the Iowa Gambling Task (IGT), a complex measure of “decision-making.” Little remains known about what other neurocognitive processes may account for variability in IGT performance among HIV+ samples or the role of procedural learning (PL) in IGT performance. A sample of 49 HIV+ individuals with a history of substance use disorders was examined to explore the relationship between IGT performance and three measures of PL: The Rotary Pursuit, Mirror Star Tracing, and Weather Prediction tasks. We found no statistically significant relationships between IGT performance and any of the PL tasks, despite finding significant correlations among the PL tasks. This pattern of results persisted when analyzing IGT performance in various ways (e.g., performance on earlier trial blocks or impairment classifications). Although other nondeclarative processes (e.g., somatic markers) may be important for IGT performance, these findings do not support PL as an important component neurocognitive process for the IGT. Similarly, these results suggest that differences in PL performance does not account for the decision-making deficits or variability in performances observed on the IGT among HIV+ individuals with a history of substance dependence.
Keywords: HIV, Substance use disorders, Nondeclarative memory, Implicit memory, Decision-making, Basal ganglia, Orbitofrontal cortex, Executive functions
HIV/AIDS is associated with damage to structures in basal ganglia, prefrontal cortex, and their white matter pathways. As such, deficits have been documented among HIV+ samples in various neuropsychological domains, including processing speed, memory, and executive functions (Heaton et al., 1995; Reger, Welsh, Razani, Martin, & Boone, 2002). More recently, impairments in “decision-making,” as assessed by the Iowa Gambling Task (IGT), have also been reported among HIV+ groups (Hardy, Hinkin, Levine, Castellon, Lam, 2006; Martin et al., 2004). The IGT simulates real-life decision-making by requiring participants to select the most advantageous responses from a set of options with an ambiguous schedule of rewards and punishments that are not shared with participants (Bechara, Damasio, Damasio, & Anderson, 1994). This task is thought to measure the broad construct of decision-making, with poor performance described as reflecting a “myopia for the future,” where choices are driven more by immediate outcomes than by long-term consequences (Bechara, Dolan, & Hindes, 2002; Bechara et al., 1994). This construct is of significant interest in the context of HIV as prior studies have suggested that performance on this task may impact risky behaviors (Gonzalez et al., 2005). Thus, performance on the IGT may have implications for substance use, risky sexual practices, and medication adherence among HIV+ individuals. Although the IGT is generally considered a measure of decision-making, it is a complex task that likely relies on numerous neurocognitive processes, yet little remains known on how other neurocognitive abilities relate to IGT performance among HIV+ individuals. For example, as we discuss in detail below, the role of nondeclarative memory in IGT performance remains controversial and in need of further study. In this investigation, we examine if deficits on the IGT among HIV+ individuals with a history of substance dependence may be explained, at least in part, by deficits in a specific aspect of nondeclarative memory (i.e., procedural learning [PL]).
PL refers to an aspect of nondeclarative memory including gradual, incremental learning of associations, skills, and habits that can be demonstrated through improvements in task performance, but do not require conscious memorization or recollection (e.g., riding a bike, tennis swing, driving). Thus, deficits in PL may affect acquisition of new skills and habits. There are reasons to suspect that HIV may adversely affect PL performance, though studies specifically examining this question have not been conclusive. The brain systems primarily implicated in PL overlap with those preferentially affected by HIV. Specifically, basal ganglia structures (especially caudate and putamen) have been consistently reported as vital for PL (Packard & Knowlton, 2002; Squire & Zola, 1996; Yin & Knowlton, 2006). Clinical populations with diseases that produce severe basal ganglia pathology (e.g., Parkinson's and Huntington's disease) perform more poorly on PL measures (Heindel, Butters, & Salmon, 1988; Salmon & Butters, 1995; Shohamy, Myers, Onlaor, & Gluck, 2004). Damage to structures of the basal ganglia and its associated circuits are also commonly reported among HIV+ individuals through a variety of neuroimaging and neuropathological methods (Aylward et al., 1993; Chang et al., 2001, 2004; Ernst, Itti, Itti, & Chang, 2000; Jernigan et al., 1993; Masliah, Ge, Achim, DeTeresa, & Wiley, 1996; Meyerhoff et al., 1999; Nath et al., 2000; Navia, Cho, Petito, & Price, 1986; Paul et al., 2007; Rottenberg et al., 1996). The neuropsychological deficits observed in those with HIV are reported to resemble those observed among patient groups with damage to basal ganglia (e.g., Huntington's and Parkinson's disease) rather than by patient groups with damage primarily to other brain systems (Sadek et al., 2004). However, it is worth noting that HIV-associated neuropathology often involves other brain structures outside of basal ganglia as well (e.g., Moore et al., 2006).
Importantly, neurocognitive dysfunction and brain abnormalities may be more severe and prevalent among HIV+ persons with substance use disorders as a result of various interacting mechanisms, such as oxidative stress, microvascular injury, and inflammatory processes (e.g., reviewed in Gonzalez & Cherner, 2008). Yet, studies of PL among HIV+ groups (with or without substance use disorders) have been scarce and those that have been conducted reveal mixed findings. For example, some studies examining PL deficits among HIV+ samples have reported significant differences on measures of PL compared with a control group (Martin, Heyes, Salazar, Law, & Williams, 1993). More recently, Gonzalez and colleagues (2008) found differences between HIV+ and HIV− individuals with a history of substance use disorders on PL tasks, but their pattern of performances suggested problems with general motor skills rather than PL deficits, per se.
Numerous investigations have reported deficits in decision-making among samples of substance users (e.g., Bechara & Damasio, 2002; Bechara et al., 2002; Gonzalez, Bechara, & Martin, 2007; Grant, Contoreggi, & London, 2000). However, to our knowledge, only two published studies have examined performance on the IGT in the context of HIV, and both reveal deficits among HIV+ groups relative to controls. Martin and colleagues (2004) compared performance on the IGT between 46 HIV+ and 47 HIV− men with a history of substance dependence and found poorer performance among the HIV+, substance-dependent men. Similar results were reported by Hardy and colleagues (2006) in a sample with fewer drug users. Substance use history was not a statistically significant covariate in their analyses, suggesting that HIV (rather than substance use alone) may account for poorer IGT performance in these samples. However, HIV+ individuals in both studies showed substantial variability in their performance on the IGT and both studies examined if performance on other neurocognitive tasks related to performance on the IGT. No significant relationships between IGT performance and working memory were reported by Martin and colleagues (2004); however, Hardy and colleagues (2006) found significant correlations between IGT performance and measures of response inhibition and declarative memory. Thus, some of the variance observed in IGT performance may have been accounted for by response inhibition and declarative memory, rather than decision-making deficits, per se. Neither study included measures of PL.
The IGT and measures of PL share many features, thus it is reasonable to hypothesize that deficits in decision-making among HIV+ substance users may in part be explained by PL deficits. For example, on both the IGT and most PL tasks, subjects are not instructed on what they must specifically do to improve their performance. This is generally “learned” by the subject as the task progresses. Subjects completing the IGT are told that they are to win as much money as possible, but are not instructed on its reward/punishment schedule. Yet, normal healthy controls show a shift in their preferences for advantageous decks as the task progresses (Bechara, Tranel, & Damasio, 2000). This is similar to the Weather Prediction Task (WPT), a task of PL in which participants are instructed to “guess” whether patterned cards presented to them predict “sunshine” or “rain,” but are not informed of the probability structure of the task. As with PL tasks, a popular method of calculating performance on the IGT is to examine improvements in performance by calculating advantageous (or disadvantageous) choices (or their net difference) across trial blocks. Thus, the variable of interest is change in performance over trial blocks rather than the absolute level of performance. This allows the investigator to examine between-group differences in the “learning” that takes place as the task progresses. Performance between healthy controls and clinical groups is often similar during the early trial blocks but diverges on the latter trial blocks. Like PL tasks, this learning does not appear to require conscious awareness.
Indeed, the development of the IGT is associated with the somatic marker hypothesis (Damasio, 1994), which posits that signals arising from bioregulatory mechanisms and the brain's representation of body states may influence responses to stimuli. Similar to the learning that takes place with PL, somatic markers may affect behavior in a “covert” (nondeclarative) manner; that is, without an individual being consciously aware of their impact. Bechara and colleagues (1997) found that healthy subjects demonstrate preferences for advantageous decks before becoming more explicitly aware of their reward/punishment schedules. On the basis of responses to verbal probes and skin conductance measures, they characterized the subjects' increasing acquisition of explicit knowledge on contingencies of the decks into four approximate stages across the 100 trials of the task: Pre-punishment (before 10th card), pre-hunch (between 10th and 50th card), hunch (between 50th and 80th card), and conceptual (after 80th card). Thus, most healthy subjects apparently rely on nondeclarative processes throughout most of the task. In further support of this idea, Stocco and Fum (2008) reported that healthy subjects developed implicit (unconscious) biases early in the task and that they lacked sufficient explicit knowledge to successfully continue making advantageous choices when told that the “good” and “bad” decks had been reversed. Another study reports that a patient with dense anterograde amnesia performed well on the IGT (implying nondeclarative processes; Turnbull & Evans, 2006); however, others have reported that most amnestic patients in their sample perform at chance on the task (Gutbrod et al., 2006). Nonetheless, it still remains unclear if PL is involved in this process.
It is important to note, however, that the role for nondeclarative processes in IGT performance has been contentious (e.g., Bechara, Damasio, Tranel, & Damasio, 2005; Maia & McClelland, 2004, 2005). Some have argued that subjects possess some explicit knowledge about outcomes on the IGT when making advantageous choices, throughout the task (Bowman, Evans, & Turnbull, 2005; Evans, Bowman, & Turnbull, 2005; Maia and McClelland, 2004). Others have suggested that discrepancies in findings may be an artifact of methodology (Persaud, McLeod, & Cowey, 2007). More recently, Gupta and colleagues (2009) examined if delay between card choices may have accounted for these discrepant findings. They found no improvements in IGT performance among five densely amnestic patients regardless of delay between cards and concluded that declarative memory (and intact midline temporal lobe structures) is critical for adequate performance on the IGT. The one study that has examined correlations between the IGT and a single PL task (the WPT) found no significant correlations among patients with Parkinson's disease, but a significant inverse correlation among healthy controls (Perretta, Pari, & Beninger, 2005).
In this investigation, we examine relationships among several measures of PL and performance on the IGT in a sample of HIV+ individuals with substance dependence. This achieves two purposes. First, we will determine if the performance of HIV+ substance-dependent individuals on the IGT is accounted for, at least in part, by their performance on measures of PL. If so, this would suggest that the observed decision-making deficits reported among HIV+ individuals may be less specific than is currently thought, since some (or all) of the variance in their performance on the IGT would be accounted for by PL. Second, the results of this study could potentially lend support to the hypothesis that the IGT is influenced by PL. Specifically, we analyze correlations between improvements in performance across trial blocks of PL tasks and improvements in performance on the IGT. We hypothesize that improvements in PL performance will be associated with improvements in IGT scores. Moreover, correlations will be strongest when considering improvements across the first three blocks of the IGT (i.e., prior to the “hunch”/“conceptual” stages of Bechara's model, which are thought to rely more heavily on nondeclarative processes) and weaker when considering improvements in performance across all five IGT blocks. Given their similarities and lack of strong motor demands, we anticipate correlations to be stronger between the WPT and the IGT when compared with correlations among the IGT and the other PL measures with stronger motor components (Mirror Star Tracing [MST] and Rotary Pursuit Task [RPT]).
Participants
Participants were 49 HIV+ individuals with a history of substance dependence recruited from the (Chicago metropolitan) community, who provided informed consent. Study procedures were in accordance with the ethical standards presented in the 1964 Declaration of Helsinki and were approved by the Institutional Review Board of the University of Illinois, Chicago. This sample is a subset of participants from a prior investigation (Gonzalez, et al., 2008). Briefly, all participants were HIV+ and met DSM-IV criteria for past cocaine and/or heroin dependence. None had a history of schizophrenia, neurological illness, less than 8 years of education, or loss of consciousness more than 30 min. Sample characteristics are summarized in Table 1. Regarding biomarkers of HIV disease severity, few participants (17%) met criteria for an immunological AIDS diagnosis, as the vast majority had CD4 T-lymphocyte counts >200 (CD4 counts: Mdn = 367, IQR [268, 513]). HIV RNA viral load in plasma was undetectable in 46% of the sample (log10 plasma viral load: Mdn = 2.44, IQR [1.88, 3.61]). Most participants (83%) were taking antiretroviral medications and 49% were on HAART.
Table 1.
Table 1.
Characteristics for sample of HIV+ individuals with history of substance dependence (n = 49)
Assessment Protocol
Participants completed two separate visits lasting about 2 hr each and consisting of structured clinical interviews, self-report questionnaires, the computer-administered IGT, and three measures of PL. Participants abstained from street drug use for at least 7 days prior to each study visit. Those testing positive on urine toxicology testing (for opiates or cocaine) or an alcohol breath test were rescheduled. Substance use diagnoses were ascertained with the Substance Abuse Module of the Structured Clinical Interview for DSM-IV (First, Spitzer, Gibbon, & Williams, 1996).
Detailed administration procedures for the IGT are presented in prior manuscripts (e.g., Gonzalez et al., 2007; Martin et al., 2004). Briefly, participants were instructed to win as much money as possible by making selections from one of four decks. Participants were not told that choices from two of the decks resulted in small monetary gains and occasionally small monetary losses (good decks), whereas the other decks yielded larger monetary gains with even larger losses (bad decks). Making more choices from bad decks results in an overall net loss across the 100 trials of the task. Choosing more cards from good decks results in overall net gains. We quantified performance by tabulating the number of choices from good decks across each of five 20-trial blocks.
Participants also completed three separate measures of PL: The photoelectric RPT, the MST, and the WPT. Detailed procedures for these tasks can be found in Gonzalez et al. (2008). The RPT, MST, and WPT are performed abnormally by patients with damage to the basal ganglia, such as those with Parkinson's and Huntington's disease compared with persons with brain disorders with relative sparing of the basal ganglia (Salmon & Butters, 1995).
The RPT requires participants to hold a plastic stylus over a rotating disk, keeping the stylus directly over a patch of light that spins around the circumference of the disk at a set speed (set to 55 rpm for all trials). Eight trials lasting 20 s each were conducted. The seconds that the stylus was kept on the target during each trial was recorded.
The MST requires participants to trace within an outline of a six-point star on a flat metallic plate using a metal stylus. Participants do not see the actual star (or their hand) as they are tracing, but rather see only a mirror image. Eight trials were conducted, with the number of seconds required for the participant to trace the full outline of the star, one time, recorded for each trial.
Participants also completed the WPT, a computer-administered, two-choice probabilistic classification task (Knowlton, Squire, & Gluck, 1994). On each trial, one, two, or three cards are displayed on the computer screen, each with a unique pattern. There are four different patterns that appear in various combinations on each trial. Each pattern is associated with a fixed probability of sunshine or rain that is unknown to participants (Experiment 2 of Gluck, Shohamy, & Myers, 2002). Participants are told they may have to “guess at first” but that they should try “to get better at predicting the weather” as the task goes on. Performance was quantified by tabulating the number of correct selections made during each of four 50-trial blocks. Healthy controls show increased basal ganglia activity on fMRI when performing the WPT (Poldrack et al., 2001) and patients with Parkinson's disease evidence deficits (Shohamy et al., 2004).
General Statistical Procedures
Distributions of data for each variable and all statistical analyses were examined for outliers and violations of statistical assumptions. In order to reduce trial-to-trial variability on the RPT and the MST, performance on the eight trials of the task were reduced to four-trial blocks for each of these two measures; that is, each trial block represented the average performance across two successive trials (i.e., Trial Block 1 = average of Trials 1 and 2; Trial Block 2 = average of Trials 3 and 4; etc.). A square-root transformation was applied to data from the final trial block (Trial 4) of the MST and MST change scores due to non-normal distributions. To reduce the risk of Type-I errors, analyses were deemed statistically significant when p-values were <.01. Results with p-values of <.10 were deemed to approach statistical significance.
There are several possible ways to examine performance and the relationships among these tasks. Our primary construct of interest is PL, which can be captured by the difference in performance from the first to the last trial block on each task. Such a “change score” would be specific to the learning taking place (as captured by improvements in performance) over the course of the task. In contrast, one could examine the relationships between initial Trial Block 1 scores, which allows the investigation of whether baseline levels of performance are correlated among tasks. For example, baseline differences in psychomotor skills might affect initial performance on both the MST and the RPT (even before any learning has taken place) and could also possibly contribute to an overall better performance on the last trial block of the task. Finally, one could examine absolute levels of attainment in performance on the task by examining only the final trial blocks, which would capture both baseline differences in performance on the task as well as any learning that takes place.
Because PL on the tasks was the construct of interest, for all tasks change scores (computed by subtracting performance on the last trial block from the first trial block) were the primary variables examined. For example, the RPT change score was the difference between average seconds on target during Trial Block 4 and Trial Block 1. Two separate change scores were computed for the IGT. The first (CS5-1) represented the difference between the average number of good choices from the last and first trial block of the IGT. The second change score (CS3-1) represented the difference between the third and first trial blocks. Thus, CS5-1 captures changes in performance across the entire IGT, whereas CS3-1 captures improvements in performance when participants are still not aware of task contingencies (i.e., prior to the Conceptual Stage, according to Bechara's model). Both change scores showed ample variability in the sample, with many participants improving in performance across trial blocks and many others performing more poorly as the task progressed (CS5-1: Range = −9 to 13; CS3-1: Range = −10 to 13). A secondary set of exploratory analyses was conducted to examine if different patterns of correlations emerged when examining initial and absolute levels of performance in comparison to the rate of learning. Higher scores represent better performance on all tasks, with the exception of the MST. Therefore, the sign of MST raw scores and change scores was reversed for our analyses. Mean level of performance on each trial block, as well as overall mean level of performance and average change scores, is presented for all tasks in Table 2.
Table 2.
Table 2.
Performance of groups on neurocognitive measures (n = 49)
Relationships Among the IGT and PL Performance
Pearson's product–moment correlations were calculated for the change scores on each task (Table 3). Change scores for the RPT were significantly correlated with the WPT (r = .42, p = .002) and showed a trend toward statistical significance with the MST (r = .26, p = .076). In contrast, MST and WPT change scores were not significantly correlated (r = −.12, p = .42). No statistically significant correlations were observed between change scores from the IGT CS5-1 and the RPT (r = .23, p = .12), MST (r = .11, p = .46), or WPT (r = −.035, p = .76). Similarly, no significant correlations were observed between change scores from any of the PL tasks and the CS3-1 score of the IGT (RPT: r = .046; p = .76; MST: r = .067; p = .65; WPT: r = −.017, p = .91).
Table 3.
Table 3.
Pearson correlations of change in performance for procedural learning measures and the Iowa Gambling Task
Owing to the medium to small correlations observed among some of the PL tasks, we conducted further analyses to examine if PL task change scores jointly accounted for significant variance in IGT change scores. Multiple linear regressions were conducted with PL task change scores as the independent variables and either CS5-1 or CS3-1 change scores on the IGT as the dependent variable. Omnibus models of both analyses revealed no statistically significant relationships (CS3-1: R2 = 0.006, p = .97; CS5-1: R2 = 0.072, p = .34).
Finally, we examined if less improvement on PL tasks (i.e., poorer PL) was associated with increased odds of obtaining an overall “impaired” performance on the IGT. Participants were classified as impaired using criteria presented by Bechara and colleagues (2001). Three separate nominal logistic regressions were conducted with PL change scores as the independent variable and IGT impairment classification as the dependent variable. None of the three models were statistically significant (all logit R2 < 0.01, all p-values >.40).
Exploratory Analyses Examining Initial and Absolute Performance
Pearson's product–moment correlations were also calculated for absolute performance scores at the final block of each PL task (Table 4). Paralleling the analyses above, PL task correlations with IGT Task performance at Block 3 (representing absolute level of performance prior to the conceptual stage) and at Block 5 (representing absolute level of performance after the conceptual stage) were calculated. The correlation pattern was remarkably similar to that found for the change scores (Table 3), with the exception that the WPT and MST, which were uncorrelated in the change score analysis (r = −.12, p = .42), were moderately correlated when absolute levels of performance were examined (r = .36, p = .01).
Table 4.
Table 4.
Pearson correlations of final block performance on procedural learning measures and Iowa Gambling Task performance
A similar set of analyses was carried out correlating only levels of performance on initial trial blocks. Pearson's product–moment correlations among initial trial blocks of all tasks were not statistically significant (all p > .05; Pearson's r range from −.07 to .24).
Prior studies have reported deficits in decision-making, as assessed by the IGT, among HIV+ individuals with and without substance use disorders (Hardy et al., 2006; Martin et al., 2004). However, HIV+ individuals in these samples show ample variability in their performance on the IGT; therefore, it is possible that deficits in abilities other than decision-making may be affecting their performance on this task. The IGT is a complex task that likely requires various component neurocognitive processes for adequate performance. The current investigation was prompted by evidence of deficits on the IGT among HIV+ individuals, coupled with a lack of knowledge on the role of other neurocognitive abilities accounting for such deficits. Furthermore, nondeclarative processes have been implicated in IGT performance and the IGT shares many features with measures of PL, but there is a dearth of studies examining their relationship. In this manuscript, we set out to examine if performance on the IGT among HIV+ individuals with substance dependence was influenced by PL. Contrary to our hypotheses, we found no significant associations between three different measures of PL and performance on the IGT. Indeed, the absence of significant relationships persisted across various sets of analyses, including those that examined initial and absolute levels of performance and those that only examined performance on the earlier trials of the IGT, when participants are thought to rely more on nondeclarative processes. Moreover, measures of PL did not predict impairment classification on the IGT. Thus, we found no evidence to suggest that PL is important for performing the IGT in this sample. In contrast, we did find significant correlations among the three measures of PL, suggesting that they do indeed measure a similar construct. The absence of significant correlations between PL tasks and the IGT were not likely the result of differences in motor demands among these tasks, as we found significant correlations between two measures of PL that vary significantly in their motor demands (i.e., the rotary pursuit and WPT); whereas the WPT and the IGT, which have very similar motor demands, were not correlated.
There are several possible reasons why we found no significant associations between PL and IGT performance in our sample. First, it is possible that differences in the sensitivity of the IGT and PL tasks to HIV-associated cognitive impairment may have affected our findings. It is important to consider that our sample consisted of HIV+ individuals in relatively good health, as indexed by measures of immune function. Specifically, almost half the sample had undetectable HIV viral load in plasma and only a fairly small number met an immunological AIDS diagnosis. If deficits in decision-making or PL manifest at differing stages of HIV disease, with one being affected sooner than the other, then it is possible that correlations that would become evident with more advanced disease would be obfuscated in this relatively healthy sample. We do note, however, that participants in our sample showed a broad range of performance across all tasks, with no evident floor or ceiling effects. However, future studies should explore correlations between PL tasks and the IGT across varying disease stages.
Another very important consideration is that the present study examined PL and IGT performance only in a sample of HIV+ individuals with a history of substance dependence. Results may be different in other populations, such as HIV+ individuals without substance dependence or among healthy controls. This is likely given findings from other studies that examined both PL and IGT performance in different patient groups and healthy controls (Beninger et al., 2003; Perretta et al., 2005). Deficits in decision-making among substance users have been well established, with numerous studies reporting poor performance on the IGT compared with controls (e.g., Bechara & Damasio, 2002; Bechara et al., 2002; Gonzalez et al., 2007; Grant et al., 2000). However, our knowledge of PL among samples of individuals with substance dependence and HIV remains very limited. Similarly, little is known about PL among those with substance dependence (cf. van Gorp et al., 1999). Although we hypothesized that PL performance would account for variance on the IGT in this sample, clearly more work remains to be conducted to better understand the interplay between these constructs. Because we examined only a group of HIV+ individuals with a history of substance dependence, we are unable to dissociate any specific impact of HIV or substance use on the observed correlations. It is worth noting that inferences on independent effects of substance use and HIV in prior studies were also somewhat limited. Of the two studies published to date examining the effects of HIV on IGT performance, one also included only individuals with a history of substance use disorders (both HIV+ and HIV−; Martin et al., 2004), whereas the other had many individuals with a history of substance use disorders among their HIV+ participants, but covaried for the effects of substance dependence in their analyses (Hardy et al., 2006). At this time, it is prudent to limit our conclusions only to HIV+ substance users. Future studies should examine PL and IGT performance across a variety of clinical and healthy samples to determine if these results generalize to other populations.
Although current evidence suggests overlap in brain systems affected by HIV and substance use disorders, correlations between PL tasks and the IGT may be affected to the extent that substance use and HIV preferentially affect different structures within basal ganglia and the respective importance of such structures for PL and decision-making. Despite having unique pharmacological properties, substances of abuse have been shown to affect the mesocorticolimbic dopaminergic system (Rogers & Robbins, 2001) and a similar network of brain structures including prefrontal cortex, anterior cingulate, hippocampus, amygdala, nucleus accumbens, and ventral tegmental area (Goldstein & Volkow, 2002). These brain systems overlap in part with those affected by HIV and studies examining combined effects of HIV and substance use suggest that they compound neurocognitive impairments (e.g., reviewed in Gonzalez & Cherner, 2008). Although basal ganglia structures are affected by both of these disorders, there is some evidence to suggest that dorsal striatal structures (e.g., caudate) may be more readily affected by HIV, whereas substance use disorders preferentially affect more ventral striatal structures (e.g., nucleus accumbens). We also note that there are substantial differences between the neuroanatomical substrates implicated in IGT performance and those thought important in PL (including the WPT). Specifically, the caudate and putamen have been consistently reported as vital for PL (Packard & Knowlton, 2002; Salmon & Butters, 1995; Squire & Zola, 1996) as well as a distributed network including prefrontal, occipital, parietal cortex, and cerebellum (Grafton et al., 1992; Poldrack et al., 2001). In contrast, investigations important in the development of the somatic marker hypothesis and those examining performance on the IGT identify a system that includes ventromedial prefrontal cortex, amygdala, insula, supplementary motor cortex, and brainstem (Bechara & Damasio, 2005). On the basis of recent findings one could speculate that mid-temporal lobe structures are also critical for IGT performance. Thus, differences in specific neural systems subserving PL and decision-making, as well as those affected by HIV and substance use disorders may also explain the lack of correlations observed in the current study.
Of studies that have more broadly examined the role of nondeclarative processes in IGT performance, perhaps the most relevant to the current study's findings are those that have examined individuals with dense anterograde amnesia. There is extensive scientific literature demonstrating that individuals with anterograde amnesia due to mid-temporal lobe damage demonstrate intact nondeclarative memory, including adequate performance on tasks of PL (Heindel et al., 1988; Knowlton, Mangels, & Squire, 1996; Knowlton et al., 1994; Squire, 1987). As such, one could reason that if intact PL was sufficient for performing well on the IGT, such patients should demonstrate adequate performance. Evidence in support of this hypothesis was presented in an initial case study by Turnbull and Evans (2006) that reported on S.L., an 85-year-old subject with profound amnesia secondary to posterior cerebral artery infarction. SL performed at a level comparable to healthy controls on the IGT despite severe impairments on measures of episodic memory. Moreover, he demonstrated evidence of persisting “knowledge” of the IGT structure over weeks. In contrast, Gutbrod and colleagues (2006) examined 11 patients with amnesia from varied etiologies (e.g., ruptured anterior communication artery aneurysms, herpes encephalitis, hypoxia, thalamic infarct) and found that only three of the amnestic patients showed evidence of reaching the hunch stage of the IGT and only one reached the conceptual stage. Thus, contrary to the findings of Turnbull and Evans, they concluded that explicit memory was necessary for adequate performance on the IGT. However, they also pointed out that their study employed a 6-s delay between subjects' card selection and feedback (in order to collect skin conductance data), whereas Turnbull and Evans provided SM with immediate feedback. This difference in methodology was speculated to possibly account for the disparate findings. Gupta and colleagues (2009) addressed this issue in a recent manuscript by administering the IGT to five amnestic patients using a 6-s delay or immediate feedback during separate administrations. Regardless of the delay condition, amnestic patients performed poorly on the IGT by not making more advantageous choices across trial blocks. All of these investigations focused on examining episodic memory and did not specifically measure PL or other aspects of nondeclarative memory in their sample. Nonetheless, when taken together with the findings of the current study, most of the evidence to date suggests that PL is neither necessary nor sufficient for performing adequately on the IGT.
Despite their similarities, the IGT and the measures of PL employed in this investigation differ in important ways that may explain the lack of correlations observed in this study. As noted in the Introduction, the IGT is closely associated with the somatic marker hypothesis, thus performance on the IGT is thought to be influenced by bioregulatory mechanisms and the brain's representation of body states, which do not require conscious awareness to influence behavior. The IGT requires participants to make a choice between obtaining large immediate rewards (at the expense of longer-term losses) or choosing smaller immediate rewards that yield better long-term outcomes and it provides subjects with simulated rewards and punishments after choices are made. The PL tasks employed in this manuscript (and typically examined in the literature) lack a reward/punishment structure. Thus, the IGT is likely to engage the limbic system and prefrontal cortex to a larger extent than PL tasks. As has been suggested by others (e.g., Beninger, 2006), we think that the “incentive learning” or “emotion-based knowledge” that takes place when performing the IGT represents a different subtype of nondeclarative memory than that assessed by PL tasks such as the WPT. The absence of incentive or emotion-based learning and simulated rewards or punishments in the PL tasks used in this study suggests that somatic markers play a less important role (or none at all) in guiding behavior on PL tasks. The current study only examined one aspect of nondeclarative memory (i.e., PL) and did not administer other emotion-based learning tasks. We cannot rule out the possibility that other nondeclarative processes play some role in performance of the IGT.
In summary, the findings of the current study do not lend support to the hypothesis that PL may be an important component of the IGT, as we found that PL does not contribute to the variability that has been observed on the IGT among HIV+ individuals. Taken together with prior research findings, it appears that decision-making deficits among HIV persons are influenced in part by response inhibition and declarative memory (Hardy et al., 2006), but not working memory or PL (Martin et al., 2004). However, none of these other cognitive constructs appear to account for most of the variance in IGT performance observed among HIV+ groups, suggesting that such deficits may indeed be the result of problems with decision-making. This is an important distinction, as deficits in decision-making may have different ramifications for other risky behaviors, such as risky sexual or drug use practices that would serve for vectors of HIV transmission or re-infection. Further research is still needed to understand if the neuropathology associated with HIV directly results in decision-making deficits, or if such deficits are mediated primarily through impairments in other cognitive abilities that have yet to be examined. It is also possible that some deficits in decision-making may precede HIV infection in this population. More broadly, when interpreted in the context of emerging research, the findings of the current study suggest that PL may not be important to the performance of the IGT.
Funding
This study was supported in part by grants from the National Institute on Drug Abuse, National Institutes of Health (F32 DA018522 and K23 DA023560 to RG, R01 DA12828 to EMM).
Conflict of interest
None declared.
Acknowledgements
The authors thank Drs Rodney Eiger and Max Brito, as well Gerald Nunnally at the Jesse Brown VA Medical Center for their cooperation in referring participants for this investigation. Special thanks to Dr Antoine Bechara for his feedback on an early draft of this manuscript.
  • Aylward E. H., Henderer J. D., McArthur J. C., Brettschneider P. D., Harris G. J., Barta P. E., et al. Reduced basal ganglia volume in HIV-1-associated dementia: Results from quantitative neuroimaging. Neurology. 1993;43(10):2099–2104. [PubMed]
  • Bechara A., Damasio H. Decision-making and addiction (part I): Impaired activation of somatic states in substance dependent individuals when pondering decisions with negative future consequences. Neuropsychologia. 2002;40(10):1675–1689. [PubMed]
  • Bechara A., Damasio A. R. The somatic marker hypothesis: A neural theory of economic decision. Games and Economic Behavior. 2005;52(2):336–372.
  • Bechara A., Damasio A. R., Damasio H., Anderson S. W. Insensitivity to future consequences following damage to human prefrontal cortex. Cognition. 1994;50(1–3):7–15. [PubMed]
  • Bechara A., Damasio H., Tranel D., Damasio A. R. Deciding advantageously before knowing the advantageous strategy. Science. 1997;275(5304):1293–1295. [PubMed]
  • Bechara A., Damasio H., Tranel D., Damasio A. R. The Iowa Gambling Task and the somatic marker hypothesis: Some questions and answers. Trends in Cognitive Sciences. 2005;9(4):159–162. [PubMed]
  • Bechara A., Dolan S., Denburg N., Hindes A., Anderson S. W., Nathan P. E. Decision-making deficits, linked to a dysfunctional ventromedial prefrontal cortex, revealed in alcohol and stimulant abusers. Neuropsychologia. 2001;39(4):376–389. [PubMed]
  • Bechara A., Dolan S., Hindes A. Decision-making and addiction (part II): Myopia for the future or hypersensitivity to reward? Neuropsychologia. 2002;40(10):1690–1705. [PubMed]
  • Bechara A., Tranel D., Damasio H. Characterization of the decision-making deficit of patients with ventromedial prefrontal cortex lesions. Brain. 2000;123(Pt. 11):2189–2202. [PubMed]
  • Beninger R. J. Dopamine and incentive learning: A framework for considering antipsychotic medication effects. Neurotoxicity Research. 2006;10:199–209. [PubMed]
  • Beninger R. J., Wasserman J., Zanibbi K., Charbonneau D., Mangels J., Beninger B. V. Typical and atypical antipsychotic medications differentially affect two nondeclarative memory tasks in schizophrenic patients: A double dissociation. Schizophrenia Research. 2003;61(2–3):281–292. [PubMed]
  • Bowman C. H., Evans C. E., Turnbull O. H. Artificial time constraints on the Iowa Gambling Task: The effects on behavioural performance and subjective experience. Brain and Cognition. 2005;57(1):21–25. [PubMed]
  • Chang L., Lee P. L., Yiannoutsos C. T., Ernst T., Marra C. M., Richards T., et al. A multicenter in vivo proton-MRS study of HIV-associated dementia and its relationship to age. NeuroImage. 2004;23(4):1336–1347. [PubMed]
  • Chang L., Speck O., Miller E. N., Braun J., Jovicich J., Koch C., et al. Neural correlates of attention and working memory deficits in HIV patients. Neurology. 2001;57(6):1001–1007. [PubMed]
  • Damasio A. R. Descartes' error. New York: G. P. Putnam; 1994.
  • Ernst T., Itti E., Itti L., Chang L. Changes in cerebral metabolism are detected prior to perfusion changes in early HIV-CMC: A coregistered (1)H MRS and SPECT study. Journal of Magnetic Resonance Imaging. 2000;12(6):859–865. [PubMed]
  • Evans C. E., Bowman C. H., Turnbull O. H. Subjective awareness on the Iowa Gambling Task: The key role of emotional experience in schizophrenia. Journal of Clinical and Experimental Neuropsychology. 2005;27(6):656–664. [PubMed]
  • First M. B., Spitzer R. L., Gibbon M., Williams J. B. Structured clinical interview for DSM-IV axis I disorders. New York: Biometrics Research Department; 1996.
  • Gluck M. A., Shohamy D., Myers C. How do people solve the “weather prediction” task?: Individual variability in strategies for probabilistic category learning. Learning and Memory. 2002;9(6):408–418. [PubMed]
  • Goldstein R. Z., Volkow N. D. Drug addiction and its underlying neurobiological basis: Neuroimaging evidence for the involvement of the frontal cortex. American Journal of Psychiatry. 2002;159(10):1642–1652. [PMC free article] [PubMed]
  • Gonzalez R., Bechara A., Martin E. M. Executive functions among individuals with methamphetamine or alcohol as drugs of choice: Preliminary observations. Journal of Clinical and Experimental Neuropsychology. 2007;29(2):155–159. [PubMed]
  • Gonzalez R., Cherner M. Co-factors in HIV neurobehavioral disturbances: Substance abuse, Hepatitis C, and aging. International Review of Psychiatry. 2008;20:49–60. [PubMed]
  • Gonzalez R., Jacobus J., Amatya A. K., Quartana P. J., Vassileva J., Martin E. M. Deficits in complex motor functions, despite no evidence of procedural learning deficits, among HIV plus individuals with history of substance dependence. Neuropsychology. 2008;22(6):776–786. [PMC free article] [PubMed]
  • Gonzalez R., Vassileva J., Bechara A., Grbesic S., Sworowski L., Novak R. M., et al. The influence of executive functions, sensation seeking, and HIV serostatus on the risky sexual practices of substance-dependent individuals. Journal of the International Neuropsychological Society. 2005;11(2):121–131. [PubMed]
  • Grafton S. T., Mazziotta J. C., Presty S., Friston K. J., Frackowiak R. S., Phelps M. E. Functional anatomy of human procedural learning determined with regional cerebral blood flow and PET. Journal of Neuroscience. 1992;12(7):2542–2548. [PubMed]
  • Grant S., Contoreggi C., London E. D. Drug abusers show impaired performance in a laboratory test of decision making. Neuropsychologia. 2000;38(8):1180–1187. [PubMed]
  • Gupta R., Duff M. C., Denburg N. L., Cohen N. J., Bechara A., Tranel D. Declarative memory is critical for sustained advantageous complex decision-making. Neuropsychologia. 2009;47(7):1686–1693. [PMC free article] [PubMed]
  • Gutbrod K., Krouzel C., Hofer H., Muri R., Perrig W., Ptak R. Decision-making in amnesia: Do advantageous decisions require conscious knowledge of previous behavioural choices? Neuropsychologia. 2006;44(8):1315–1324. [PubMed]
  • Hardy D. J., Hinkin C. H., Levine A. J., Castellon S. A., Lam M. N. Risky decision making assessed with the gambling task in adults with HIV. Neuropsychology. 2006;20(3):355–360. [PMC free article] [PubMed]
  • Heaton R. K., Grant I., Butters N., White D. A., Kirson D., Atkinson J. H., et al. The HNRC 500–neuropsychology of HIV infection at different disease stages. HIV Neurobehavioral Research Center. Journal of the International Neuropsychological Society. 1995;1(3):231–251. [PubMed]
  • Heindel W. C., Butters N., Salmon D. P. Impaired learning of a motor skill in patients with Huntington's disease. Behavioral Neuroscience. 1988;102(1):141–147. [PubMed]
  • Jernigan T. L., Archibald S., Hesselink J. R., Atkinson J. H., Velin R. A., McCutchan J. A., et al. Magnetic resonance imaging morphometric analysis of cerebral volume loss in human immunodeficiency virus infection. The HNRC Group. Archives of Neurology. 1993;50(3):250–255. [PubMed]
  • Knowlton B. J., Mangels J. A., Squire L. R. A neostriatal habit learning system in humans. Science. 1996;273(5280):1399–1402. [PubMed]
  • Knowlton B. J., Squire L. R., Gluck M. A. Probabilistic classification learning in amnesia. Learning and Memory. 1994;1(2):106–120. [PubMed]
  • Maia T. V., McClelland J. L. A reexamination of the evidence for the somatic marker hypothesis: What participants really know in the Iowa gambling task. Proceedings of the National Academy of Science of the USA. 2004;101(45):16075–16080. [PubMed]
  • Maia T. V., McClelland J. L. The somatic marker hypothesis: Still many questions but no answers Response to Bechara et al. Trends in Cognitive Sciences. 2005;9(4):162–164.
  • Martin A., Heyes M. P., Salazar A. M., Law W. A., Williams J. Impaired motor-skill learning, slowed reaction time, and elevated cerebrospinal fluid quinolinic acid in a subgroup of HIV-infected individuals. Neuropsychology. 1993;7(2):149–157.
  • Martin E. M., Pitrak D. L., Weddington W., Rains N. A., Nunnally G., Nixon H., et al. Cognitive impulsivity and HIV serostatus in substance dependent males. Journal of the International Neuropsychological Society. 2004;10(7):931–938. [PubMed]
  • Masliah E., Ge N., Achim C. L., DeTeresa R., Wiley C. A. Patterns of neurodegeneration in HIV encephalitis. Journal of Neuro-AIDS. 1996;1(1):161–173. [PubMed]
  • Meyerhoff D. J., Bloomer C., Cardenas V., Norman D., Weiner M. W., Fein G. Elevated subcortical choline metabolites in cognitively and clinically asymptomatic HIV+ patients. Neurology. 1999;52(5):995–1003. [PubMed]
  • Moore D. J., Masliah E., Rippeth J. D., Gonzalez R., Carey C. L., Cherner M., et al. Cortical and subcortical neurodegeneration is associated with HIV neurocognitive impairment. Aids. 2006;20(6):879–887. [PubMed]
  • Nath A., Anderson C., Jones M., Maragos W., Booze R., Mactutus C., et al. Neurotoxicity and dysfunction of dopaminergic systems associated with AIDS dementia. Journal of Psychopharmacology. 2000;14(3):222–227. [PubMed]
  • Navia B. A., Cho E. S., Petito C. K., Price R. W. The AIDS dementia complex: II. Neuropathology. Annals of Neurology. 1986;19(6):525–535. [PubMed]
  • Packard M. G., Knowlton B. J. Learning and memory functions of the Basal Ganglia. Annual Review of Neuroscience. 2002;25:563–593. [PubMed]
  • Paul R. H., Yiannoutsos C. T., Miller E. N., Chang L., Marra C. M., Schifitto G., et al. Proton MRS and neuropsychological correlates in AIDS dementia complex: Evidence of subcortical specificity. Journal of Neuropsychiatry and Clinical Neurosciences. 2007;19(3):283–292. [PubMed]
  • Perretta J. G., Pari G., Beninger R. J. Effects of Parkinson disease on two putative nondeclarative learning tasks: Probabilistic classification and gambling. Cognitive and Behavioral Neurology. 2005;18(4):185–192. [PubMed]
  • Persaud N., McLeod P., Cowey A. Post-decision wagering objectively measures awareness. Nature Neuroscience. 2007;10(2):257–261. [PubMed]
  • Poldrack R. A., Clark J., Pare-Blagoev E. J., Shohamy D., Creso M. J., Myers C., et al. Interactive memory systems in the human brain. Nature. 2001;414(6863):546–550. [PubMed]
  • Reger M., Welsh R., Razani J., Martin D. J., Boone K. B. A meta-analysis of the neuropsychological sequelae of HIV infection. Journal of the International Neuropsychological Society. 2002;8(3):410–424. [PubMed]
  • Rogers R. D., Robbins T. W. Investigating the neurocognitive deficits associated with chronic drug misuse. Current Opinion in Neurobiology. 2001;11(2):250–257. [PubMed]
  • Rottenberg D. A., Sidtis J. J., Strother S. C., Schaper K. A., Anderson J. R., Nelson M. J., et al. Abnormal cerebral glucose metabolism in HIV-1 seropositive subjects with and without dementia. Journal of Nuclear Medicine. 1996;37(7):1133–1141. [PubMed]
  • Sadek J. R., Johnson S. A., White D. A., Salmon D. P., Taylor K. I., Delapena J. H., et al. Retrograde amnesia in dementia: Comparison of HIV-associated dementia, Alzheimer's disease, and Huntington's disease. Neuropsychology. 2004;18(4):692–699. [PubMed]
  • Salmon D. P., Butters N. Neurobiology of skill and habit learning. Current Opinion in Neurobiology. 1995;5(2):184–190. [PubMed]
  • Shohamy D., Myers C. E., Onlaor S., Gluck M. A. Role of the basal ganglia in category learning: How do patients with Parkinson's disease learn? Behavioral Neuroscience. 2004;118(4):676–686. [PubMed]
  • Squire L. R. The organization and neural substrates of human memory. International Journal of Neurology. 1987;21–22:218–222. [PubMed]
  • Squire L. R., Zola S. M. Structure and function of declarative and nondeclarative memory systems. Proceedings of the National Academy of Science of the USA. 1996;93(24):13515–13522. [PubMed]
  • Stocco A., Fum D. Implicit emotional biases in decision making: The case of the Iowa Gambling Task. Brain and Cognition. 2008;66(3):253–259. [PubMed]
  • Turnbull O. H., Evans C. E. Preserved complex emotion-based learning in amnesia. Neuropsychologia. 2006;44(2):300–306. [PubMed]
  • van Gorp W. G., Wilkins J. N., Hinkin C. H., Moore L. H., Hull J., Horner M. D., et al. Declarative and procedural memory functioning in abstinent cocaine abusers. Archives of Genetic Psychiatry. 1999;56(1):85–89. [PubMed]
  • Yin H. H., Knowlton B. J. The role of the basal ganglia in habit formation. Nature Reviews: Neuroscience. 2006;7(6):464–476. [PubMed]
Articles from Archives of Clinical Neuropsychology are provided here courtesy of
Oxford University Press