Whereas the effects of smoking abstinence on neurocognition have been studied for more than 30 years, the precise causes for these effects remain to be fully elucidated. Animal studies—which have not been the focus of the current review—have provided data on the effects of nicotine withdrawal on brain dopamine and other neurotransmitter systems that subserve neurocognition. Studies have demonstrated, for instance, that cessation of nicotine administration results in substantial decreases in striatal dopamine levels (Domino and Tsukada 2009; Rahman et al. 2004), due in part to upregulated dopamine reuptake associated with nicotine administration (Fowler et al. 1996a, b). However, even within the animal literature studies linking abstinence-induced changes in neurotransmitter systems to changes in cognitive functions including learning, studies on memory and attention are rare (see Levin 2006, p. 458 for a review of this literature). There are even fewer studies that have sought to link withdrawal-induced changes in neurocognition to variability in nicotine self-administration, reinforcement, or relapse-like behavior.

Human research on the causes of abstinence-induced changes in neurocognition is similarly lacking. Whereas fMRI studies have begun to elucidate the brain circuits involved in changes in neurocognition, we are only now beginning to converge on enough data for meta-analyses of these effects. A recent meta-analysis of cue-reactivity, for instance, was able to identify only 11 fMRI studies of cue-reactivity that shared roughly comparable cue exposure paradigms and conducted whole-brain analyses allowing for pooled effects, and only a subset of those examined the effects of smoking abstinence. No such analyses have been carried out for other forms of cognition including working memory or inhibitory control and the variability in research methods, tasks, and measures will likely hamper such efforts even when sufficiently large numbers of studies are available.

Beyond fMRI, relatively little human research has sought to identify mechanisms that underlie abstinence-induced changes in neurocognition. Although PET studies have identified changes in nicotinic and other neuroreceptor availability (see chapter entitled Imaging Tobacco Smoking with PET and SPECT; volume 24), these studies have not typically sought associations between changes in smoking state and changes in cognitive performance. A small number of studies have sought to relate genetic markers associated with variability in neurotransmitter function to abstinence-induced changes in neurocognition (Loughead et al. 2009), and more work of this type—especially if it could also include direct assessment of neurochemical changes—is warranted.

Despite the research presented above showing that smoking abstinence modulates neurocognition, surprising little is known regarding the specific role of these deficits in relapse. Only a small number of studies have examined the effects of abstinence on neurocognition for more than 24 h following cessation. Studies that have looked beyond this 24 h window (Bradstreet et al. 2014; Gilbert et al. 2004) have observed that deficits in cognitive performance and brain function resolve only after prolonged abstinence (i.e., after one month). In contrast, studies of prolonged abstinence have provided some evidence that cue-reactivity remains stable or increases over the course of abstinence (Bedi et al. 2011). Despite these suggestive studies, more research is needed to establish the time course of neurocognitive changes across various domains using modern neuroimaging techniques. The differential effects of nicotine replacement and other therapies on neurocognitive trajectories are similarly lacking, but might provide clues as to the differential efficacy of interventions and could suggest ways in which to combine therapies to more comprehensively address multiple forms of neurocognition.

Likewise, more studies attempting to predict smoking cessation outcomes from either baseline or abstinence-induced changes in neurocognition are needed. To date, only a small number of such studies have been published, with most of these focused on cue-reactivity. Studies seeking to predict smoking cessation outcomes from cue-reactivity measures have produced mixed and often counterintuitive effects (Conklin et al. 2012; Perkins 2012; Powell et al. 2010; Wray et al. 2013). In one study, for instance, greater self-reported cue-reactivity was associated with an increased likelihood of initiating a quit attempt (Conklin et al. 2012); in another, greater pre-quit brain reactivity to smoking cues was associated with better cessation outcomes (McClernon et al. 2007). Another small-scale study, however, observed increased brain cue-reactivity among future relapsers, including in the anterior insula (Janes et al. 2010). Other studies have examined other neurocognitive predictors of cessation outcomes. Patterson and colleagues found that deficits in working memory following 3 days of abstinence were predictive of poorer smoking cessation outcomes among untreated smokers, while no association was seen among smokers treated with varenicline (Patterson et al. 2010). Another study found that, among smokers making an unaided quit attempt, deficits on measures of response inhibition and increased cue-reactivity were predictive of abstinence rates at 1 week, 1 month, and 3 months post quit (Powell et al. 2010). Finally, our own work has investigated abstinence-induced decrements in striatal activation to monetary rewards as a predictor of cessation outcomes (Sweitzer et al. in preparation). In that study, smokers were scanned at two time points, once after smoking as usual and once after 24 h of abstinence; they then completed a 3-week quit attempt supported by contingency management. Smokers who lapsed during the quit attempt exhibited significant decreases in striatal activation to monetary reward during abstinence relative to satiation, while those who maintained continuous abstinence exhibited no change. These results provide preliminary evidence for the importance of neurocognitive deficits in predicting cessation outcomes. However, much more work is clearly needed before definitive conclusions can be drawn.

Finally, in addition to the gaps identified above, a larger and potentially more important gap exists in our understanding of the specific role of cognitive factors in the sequence of events and behaviors leading up to lapse and relapse. Reasons for this gap include several factors: (a) lapse-like behavior can be difficult to model in the laboratory; (b) assessing neurocognition in real-time leading up to smoking lapse is challenging in the context of laboratory and real-world experiments, and (c) relapse, as a phenomenon, is not conducive to repeatability (i.e., smokers may not be keen on quitting over and over again so we can study relapse). For these reasons, whereas it is well established that abstinence results in deficits in executive function, is not known if or whether these deficits play a role in lapse behavior, either directly (negative reinforcement) or indirectly (goal and skill interference). In order to answer these questions, better models of lapse and relapse are needed for conducting controlled studies (McKee 2009; Sweitzer et al. 2013); and new methods for assessing real-time cognition and neurophysiology leading up to real-world lapses are needed.

If abstinence-induced changes in neurocognition contribute to smoking lapse and relapse, then interventions to counter or diminish these changes can potentially improve smoking cessation outcomes. As presented above, abstinence from nicotine appears to account for deficits in sustained attention, since administration of nicotine to smokers reverses these deficits (Parrott and Roberts 1991). As such, if deficits in these forms of neurocognition are linked—either directly or indirectly—to lapse and relapse, it will become vitally important that nicotine replacement therapy continues to be incorporated in smoking cessation interventions. Moreover, this may also mean that greater emphasis needs to be placed on achieving nicotine replacement at levels that replicate nicotine administration during smoking in order to achieve minimal disruption of neurocognition. This also means that other forms of procognitive therapies—whether behavioral (i.e., cognitive training; Lancaster and Stead 2005) or pharmacological (Cahill et al. 2014)—will need to demonstrate effectiveness above and beyond the effects of nicotine replacement, which remains relatively inexpensive, widely available, and with a favorable side-effect profile. In addition and related to this, additional work is needed to evaluate exactly which forms of neurocognition are remediated by nicotine and which are not. Efforts to modulate those that cannot be addressed with current first-line pharmacotherapies (nicotine, varenicline, bupropion) ought to receive the greatest attention in terms of treatment development and evaluation.