Increasing the effectiveness of decisional enhancement strategies

Several types of tobacco control DE strategies based on the principles of behavioural economics have been attempted, with varying degrees of effectiveness (Loewenstein et al., Reference Loewenstein, Asch, Friedman, Melichar and Volpp2012). As with many prevalent public health problems, even relatively small changes in behaviour (e.g. in usage rates) can have a large absolute impact on population health and societal cost. Thus, the more we understand about the neuroscience of decision making – whether it is considered to be complementary to, or to directly inform, behavioural economics (Jones, Reference Jamal, Homa, O'Connor, Babb, Caraballo, Singh, Hu and King2014; Grüne-Yanoff, Reference Grüne-Yanoff2016) – and, in particular, how the brain works in the context of addictive behaviours, the more we can tailor policy interventions to maximise effectiveness (Figure 1(b)) (Felsen & Reiner, Reference Felsen and Reiner2015).

Evidence from human and animal model studies points to an important role for the medial prefrontal cortex in mediating emotional influences on decision making and in representing risk and reward (Euston et al., Reference Euston, Gruber and McNaughton2012). This region also exhibits aberrant activity during addiction-related behaviours, including during the response to nicotine exposure, suggesting its involvement in smoking-related decisions (Cohen & George, Reference Cohen and George2013). These studies indicate that tobacco control strategies designed to modulate information processing in the medial prefrontal cortex, perhaps via eliciting emotional responses, may represent a promising approach. Indeed, warning labels that contain graphic images (e.g. a diseased lung) have been associated with thought processes linked to quitting cigarettes (e.g. perceived harm, self-reported desire to quit) (Noar et al., Reference Noar, Hall, Francis, Ribisl, Pepper and Brewer2015), and several studies suggest that graphic warning labels decrease population-wide smoking activity (Azagba & Sharaf, Reference Azagba and Sharaf2013; Jung, Reference Jung2016) and increase quit attempts and successes for active smokers (Brewer et al., Reference Brewer, Hall, Noar, Parada, Stein-Seroussi, Bach, Hanley and Ribisl2016).

A complementary strategy is suggested by evidence from neuroscience demonstrating an interesting quirk in how we update our beliefs in the face of new information. According to the ‘optimism bias’, individuals tend to change their beliefs more after being presented with favourable than unfavourable information (Sharot et al., Reference Sharot, Korn and Dolan2011). Functional magnetic resonance imaging (fMRI) studies, which measure blood flow as a proxy for neural activity, have shown that this asymmetry occurs not during deliberation, but as soon as the information is presented (Sharot et al., Reference Sharot, Korn and Dolan2011). As Tali Sharot has noted (Sharot, Reference Sharot2012), the timing of the neural response, as revealed by these fMRI studies, suggests that the information itself, rather than how it is integrated into existing beliefs, is critical in determining how the beliefs are updated. These results suggest that smokers may be more receptive to positive messages (e.g. ‘50% of smokers can quit’) than negative messages (e.g. ‘smokers die younger’), although this prediction remains to be tested. Increased dialogue between policy makers and neuroscientists would facilitate the integration of this sort of neuroscientific evidence – currently available and resulting from future studies – into the design of more effective DE strategies.

Improving the efficiency of screening decisional enhancement strategies

Once a set of potentially effective candidate DE strategies is designed, it is necessary to screen them to determine which will be most effective. Although the ultimate goal of such policies is, of course, to modify behaviour, testing multiple candidate policies at the behavioural level is inefficient. Functional neuroimaging offers the potential to streamline the process for determining which DE strategies are most likely to be effective (Figure 1(b)). Recent studies have shown that functional neuroimaging data provide additional predictive information, beyond self-reports and behavioural measures, about how individuals will respond to public health policies (Gabrieli et al., Reference Gabrieli, Ghosh and Whitfield-Gabrieli2015). Thus, a large set of policies can be screened by obtaining such data from a representative sample of participants, and only those policies most likely to be successful would then need to be tested behaviourally. Importantly, the predictive power of neuroimaging data in this context is independent of whether the data address how a particular policy exerts its effect. Returning to the medial prefrontal cortex, recent work has demonstrated that activity in this region, as well as other regions associated with emotional memory, respond differently to graphic and neutral warning labels, providing physiological evidence that is complementary to self-reports of the emotional salience of the labels (Wang et al., Reference Wang, Lowen, Romer, Giorno and Langleben2015). These findings, in turn, may explain why activity in the medial prefrontal cortex can be used to predict the success of tobacco control campaigns (Falk et al., Reference Falk, O'Donnell, Tompson, Gonzalez, Cin, Strecher, Cummings and An2016). Thus, while inexpensive approaches can continue to be used to efficiently develop DE strategies (e.g. by examining public opinion via online surveys), neuroimaging screens can complement these approaches by providing unique insights into the mechanisms by which DE strategies may work (Whelan et al., Reference Whelan, Morgan, Sherar, Orme and Esliger2017). Just as ‘neuromarketers’ have begun to use neuroimaging to efficiently identify effective advertising campaigns (Karmarkar & Yoon, Reference Karmarkar and Yoon2016), regulators charged with protecting consumers should consider a similar approach in order to identify effective DE strategies.

Informing the ethics of decisional enhancement strategies

The proposition that an individual's decision can be ‘enhanced’ raises immediate ethical questions about who is doing the enhancing and whether they have the individual's best interests in mind. We focus here on how neuroscience can inform one of the primary ethical objections to DE strategies, particularly to nudges operating below the level of conscious awareness: that they infringe upon individual autonomy (Figure 1(b)) (Bovens, Reference Bovens, Grüne-Yanoff and Hansson2009; Blumenthal-Barby & Burroughs, Reference Blumenthal-Barby and Burroughs2012; Grüne-Yanoff, Reference Grüne-Yanoff2012; White, Reference White2013; Sunstein, Reference Sunstein2015) (note that we do not consider here other important ethical issues relevant to DE strategies, such as justice or societal equity). Briefly, we suggest that evidence from neuroscience can inform this ethical concern about autonomy in two ways.

First, it can illuminate our understanding of autonomy directly, in ways that are relevant to the concern over autonomy infringement. In particular, in support of behavioural studies, neuroscience has called into question the extent to which many of our everyday decisions are sufficiently free of undue external influence to be considered autonomous in the traditional individualist sense (Custers & Aarts, Reference Custers and Aarts2010; Felsen & Reiner, Reference Felsen and Reiner2011; Bode et al., Reference Bode, Murawski, Soon, Bode, Stahl and Smith2014). To take just one example, activity in regions known to be activated by fear underlies aversion to losses, and does so covertly (De Martino et al., Reference De Martino, Kumaran, Seymour and Dolan2006; Tom et al., Reference Tom, Fox, Trepel and Poldrack2007). Instead, the evidence from neuroscience is more consistent with relational conceptions of autonomy that recognise that our decisions are, by nature, influenced by a variety of sources (Christman, Reference Christman2009; Felsen & Reiner, Reference Felsen and Reiner2011). In this case, a tobacco control DE strategy (e.g. a graphic warning label) simply provides yet one more influence on a smoking-related decision that is already influenced by numerous other sources (e.g. advertisements associating smoking with attractiveness). Thus, while the influence of DE strategies might appear to infringe upon autonomy as described by individualistic accounts, seen in the light of relational autonomy – with which the neuroscientific evidence is more consistent – these influences appears no more ethically problematic than the advertising campaigns that they are designed to counteract.

Second, neuroscience can address autonomy-related concerns about DE strategies by explicating the mechanisms by which the strategies themselves may influence decision making. In particular, some DE strategies are designed to elicit decisions consistent with ‘higher-order desires’ (e.g. staying healthy) (Frankfurt, Reference Frankfurt1971; Dworkin, Reference Dworkin1988) by counteracting detrimental internally generated biases (e.g. biological cravings), and therefore may promote, rather than infringe upon, autonomy. Although the ethics of influencing decisions is complex, accounting for the perspective from neuroscience would allow the impact of DE strategies on autonomy to be more comprehensively considered.

Limitations of neuroscience evidence for informing decisional enhancement strategies

Our goal in this paper is to explain how neuroscience can complement other disciplines in informing policies designed to improve decision making. It is also important, however, to note several limitations of using neuroscientific evidence for this purpose. First, while rapid advances in neuroscience hold great promise for informing policy in the near future, the real impact of the field is currently limited by how much we still do not understand about fundamental brain function, complicating translation to the policy domain (Seymour & Vlaev, Reference Seymour and Vlaev2012). It is therefore critical that the current value of neuroscience for policy not be overstated, particularly since neuroscientific evidence is often particularly persuasive, even if it is not entirely understood (Weisberg et al., Reference Weisberg, Keil, Goodstein, Rawson and Gray2007; O'Connor et al., Reference O'Connor, Rees and Joffe2012). In addition, cost-effectiveness should be considered: it would be impractical to replace less expensive means of data collection with costly fMRI experiments. Finally, while we have focused here on the benefits of neuroscience for informing DE strategies, we recognise that the ultimate goal of such strategies is to modify behaviour. Thus, we do not suggest that neuroscience can or should replace behavioural experiments, which will remain critical for validating DE strategies.