Cocaine addicted individuals rated drug-related words as more negative than matched neutral words, challenging the commonly held belief - but rarely tested as an empirical hypothesis in human studies - that drug cues are positively valued by drug addicted individuals. These results are consistent with a recent study where alcohol drinkers rated liking ethanol less than apple juice and water (Hobbs et al., 2005
) and implicate in the least ambivalence towards drug cues especially for the cocaine addicted individuals with histories of longer abstinence periods.
Consistent with prior reports (Whalen et al., 1998
; Bush et al., 2000
), this newly developed color-word drug Stroop fMRI task produced bilateral activations in the cdACC, in other frontal regions, parietal and occipital lobes, and in the caudate, thalamus, and cerebellum. Task hypoactivations were noted bilaterally in the rACC/mOFC, insula, parahippocampal gyrus, and in the cuneus and lingual gyrus. A trend for larger rACC/mOFC hypoactivations was noted for the drug as compared to the matched neutral words in this sample of 14 cocaine addicted individuals. This trend remains to be replicated in studies with larger sample sizes.
Further, two correlations with the prefrontal cortex were unique to the task context in this drug addicted sample: (1) the more the rACC/mOFC hypoactivation specifically to drug vs. neutral words (drug more than neutral), the higher was the number of errors for the drug vs. neutral words (drug more than neutral). Thus, although - by design (see Supplementary Material) - mean performance did not differ between the two task conditions, erroneous performance correlated with hypoactivation in the rACC/mOFC and this effect was context specific (i.e., observed only when subtracting neutral from drug responses) (); and (2) the more the cdACC activation specifically to the drug (and not neutral) words, the more negative were the valence ratings attributed to the drug (and not neutral) words (). Note that our subsequent interpretations are therefore based on these significant brain-behavior correlations, not to be attributed to ‘reverse-inference’ (Poldrack, 2006
The “affective” rACC has been previously implicated in the assessment of the salience of emotional information and regulation of emotional responses (Bush et al., 2000
), and particularly in the suppression of task-irrelevant emotional information (Whalen et al., 1998
), consistent with the mOFC role in reward processing (Elliott et al., 2000
). The first correlation may therefore be consistent with the effect of emotional context on inhibitory control within the rACC/mOFC (e.g., see (Shafritz et al., 2006
)): it is possible that the less the accuracy (more distraction) for the drug as compared to the neutral words, the greater the required rACC/mOFC suppression of an emotional reaction to the drug words in these cocaine addicted individuals. This finding (as is the lack of significant accuracy differences between the word conditions) is also consistent with the previously documented ACC error-related hypoactivity in cocaine users and other drug using groups (Kaufman et al., 2003
; Forman et al., 2004
; London et al., 2005
The “cognitive” cdACC has been previously implicated in demanding tasks that involve stimulus-response selection in the face of competing streams of information including Stroop-like tasks (Bush et al., 2000
), and particularly in conflict resolution (Carter et al., 1998
), encompassing emotionally valenced tasks (Davis et al., 2005
). The second correlation may therefore be consistent with the demanding cognitive processes that demarcate drug cues as negatively valenced for drug addicted individuals (e.g., overcoming the emotional load of drug words in order to assign what is perceived to be the subjectively appropriate or socially sanctioned negative valence).
Limitations of the current report include the following: (1) to adapt the emotional color-word Stroop task to fMRI, our design has deliberately removed the opportunity to test whether drug-related stimuli are processed behaviorally in a different way than neutral stimuli (see Supplementary Material for a detailed explanation; in brief, word reading was allowed for a long period and was separated from pressing for its color minimizing the conflict inherent in such a task. Further, display of the response key minimized working memory demands). The adaptation to the MRI environment of this color-word drug Stroop
task, achieved here for the first time, took precedence when designing the current task given the previous consistent documentation of the role for attentional bias to drug-related stimuli in drug addicted individuals; in these other word or picture color-naming or button pressing behavioral paradigms, accuracy or response latency differences between drug-related and neutral matched stimuli have been reported in abusers of alcohol (Duka and Townshend, 2004
) nicotine (Mogg and Bradley, 2002
), heroin (Franken et al., 2000
), and cocaine (Hester et al., 2005
). Nevertheless, the behavioral documentation of an attentional bias using our own task design is warranted; (2) a related concern is the lack of significant differences between the drug and neutral conditions at the BOLD signal level. We postulate that such differences would be significant with larger sample sizes or when including a non drug addicted control group; in the latter case we would expect to document a significant condition by group interaction effect on the BOLD signal in the selected ROIs such that differences between the task conditions will be observed only in the drug using and not control group (e.g., larger hypoactivation in the rACC/mOFC to the drug than neutral condition in the drug users, as indeed suggested by preliminary paired t-test ROI results in the current study); and (3) use of alternating stimuli (Cox et al., 2001
) instead of the current block design may have facilitated our ability to obtain significant behavioral differences between the drug and neutral conditions. This could be tested with further optimizations of the current design (e.g., rendering it event-related or non-verbal).
Overall, using a newly designed color-word drug Stroop fMRI task in 14 individuals with cocaine use disorders, results showed unique (drug-related) brain-behavior associations at the individual level. Thus, the more the cdACC activation and rACC/mOFC hypoactivation, the more negative the valence attributed and the more the errors committed specifically in the drug context
(controlling for the neutral context), respectively. The significance of these drug-specific multi-modal associations is in suggesting that the rACC/mOFC and the cdACC may uniquely contribute to different aspects of drug-related responses (distraction/decreased task accuracy) and intrusive negative cue-related
thoughts in drug addicted individuals. Taken together, if replicated in future studies with larger sample sizes, these results may model the on-line experience of drug addicted individuals when facing drug cues where, despite considerable amount of cognitive effort (modulated by cdACC), unregulated emotional reactions to these salient cues are paramount and are not successfully suppressed (modulated by rACC/mOFC). The effect on results of severity and duration of drug use, and application to populations with other drug use disorders, remain to be accomplished. Prospective studies also need to explore the relevance of these findings to the progression of drug addiction and prediction of relapse (see for example use of fMRI during a decision-making task to predict relapse in methamphetamine dependent subjects (Paulus et al., 2005