In this paper, I rebut Alex Voorhoeve’s (2013) argument that excise taxes on tobacco, alcohol and sugary drinks can appeal to soft paternalism in certain cases. Firstly, I outline Matthew Rabin’s (2013) case for paternalistic intervention on the grounds that irrational biases lead to self-harming habits. I then analyse Voorhoeve’s argument that, under circumstances where autonomy-enhancing strategies are ineffective, excise taxes have a soft paternalistic justification. Soft paternalistic policies constrain self-harming actions without consent, only when the action is ‘substantially non-voluntary’ (Feinberg 1986: 12). I oppose Voorhoeve’s argument by appealing to a distinction between ‘partially voluntary’ decisions and ‘substantially non-voluntary’ decisions, leading to the conclusion that excise taxes do not have a soft paternalistic justification.

Rabin (2013) analyses a behavioural bias known as projection bias. Projection bias is the tendency to incorrectly predict the future utility of an activity by adjusting it in the direction of the current utility of that activity (Loewenstein et al. 2003). This is illustrated in Figure 1, in which t=1, etc. are periods, and the numbers are within-period utilities of various actions, both perceived and real. For example, when a consumer is calculating the utility of drinking Coca-Cola at t=1, because they enjoy it greatly at t=1, projection bias causes them to upwardly bias their utility estimates of drinking Coca-Cola at t=2 and t=3. This results in an overestimation of the total utility of drinking Coca-Cola across the three periods.

Rabin argues that projection bias is irrational and causes self-harm in the presence of bad habit-forming activities (Rabin 2013). He appeals to Becker and Murphy’s (1988) definition of an activity as habit-forming if the present marginal expected utility of consuming a good is higher when an individual has consumed more of it in the past. A habit is bad if and only if past and current consumption has the additional effect of lowering future expected utility levels. By Rabin’s and Becker and Murphy’s definitions, drinking Coca-Cola is a bad habit because drinking Coca-Cola at t=1 increases the within-period marginal utility (MU) of drinking Coca-Cola at t=2, while lowering the utility level at t=2 both of drinking Coca-Cola and giving up in that period.

Figure 1: Projection bias and habit-forming activities (adapted from Voorhoeve 2013: 143). Figure 1: Projection bias and habit-forming activities (adapted from Voorhoeve 2013: 143).

Rabin argues that we consume products that lower our expected utility because of projection bias. We then continue to consume these products because they are habit forming. This results in self-harm. Rabin concludes that governments should prevent this harm by taxing behaviour that leads to unhealthy habits, referred to as excise or sin taxes (Rabin 2013: 137).

This proposal invites the charge of paternalism. Paternalistic policies limit an individual’s choices against their will. A policy is paternalistic if the purpose of the policy is to benefit the constrained individual, and cannot be fully justified without counting the benefits to those constrained (Arneson 2005). Governments are reluctant to implement paternalistic policies because these policies have many objections. One of the prominent concerns with paternalism is that it undermines the goods of autonomy. The goods of autonomy are the benefits individuals derive from the ability to live their lives according to their own independent and authentic desires, free from distorting external forces (Christman 2012). Aside from being a good in and of itself, it is argued that autonomy facilitates moral obligation and responsibility (Dworkin 1988: 121–29). Moreover, Mill (1991 [1859]) famously argued that individual autonomy is necessary to achieve human excellence. Concepts of autonomy focus on a propensity for self-rule or self-governance. The idea of self-rule contains two components: the independence of one's desires and choices from manipulation by others, and the capacity to rule oneself (Dworkin 1988: 61; Arneson 2005). By limiting individual free choice, paternalistic policies reduce autonomy since they either distort our authentic desires or prevent us from acting upon them.

Soft paternalistic policies similarly constrain self-harming actions without consent, but only do so “when conduct is substantially non-voluntary” (Feinberg 1986: 12). Voorhoeve argues that actions are substantially non-voluntary when agents are “insufficiently informed or inadequately capable of rational self-governance” (Voorhoeve 2013: 141). By limiting only non-voluntary actions, soft paternalistic policies are considerably less contentious since the goods of autonomy are not diminished. In this way, soft paternalism balances a concern for individual well-being with a respect for autonomy. Voorhoeve (2013) also argues that some excise taxes can appeal to soft paternalism if autonomy enhancing techniques are not effective at combatting projection bias. If projection bias can be corrected through autonomy enhancing techniques, this is preferable to taxation since it improves consumer autonomy rather than limiting it. For example, if the cause of projection bias is a lack of information (evidence on the declining utility of Coca-Cola consumption over time), then the best policy for well-being would be education. This policy would reduce ‘irrational’ action and enhance autonomy (Voorhoeve 2013: 142).

However, Rabin (2013) and Voorhoeve (2013) argue that some forms of projection bias cannot be effectively corrected through autonomy enhancing techniques. For example, projection bias can still be prevalent when individuals are adequately informed. Rabin refers to several studies which show that consumers’ current preferences influence their judgements of the future despite the fact that consumers possess abundant evidence that their tastes will change (Read and Van Leeuwen 1998; Lowenstein et al. 2003). If consumers are still subject to projection bias despite being sufficiently informed, then their actions could be classified as substantially non-voluntary. One explanation of consumer behaviour in these circumstances is that projection bias is the result of decision-making by the intuitive system (Kahneman 2002). The decisions of this system are fast, automatic, and easy. The intuitive system is often contrasted with the reasoning system whose decisions are slow, deliberate, controlled and effortful. Intuitive decisions are strongly influenced by current stimuli meaning that projection bias could be the result of reasoning faults by this system. Consequently, Voorhoeve (2013: 144) argues that these intuitive system decisions are substantially non-voluntary.

Voorhoeve (2013: 144) justifies his appeal to soft paternalism by arguing that intuitive system decisions cannot be corrected with autonomy enhancing techniques. Kahneman (2002) notes that we rely on the intuitive system for a variety of decisions because we have limited cognitive resources which must be used sparingly. It is therefore sensible and necessary to make many decisions in an automated and quick manner. Consequently, it would not be prudent to encourage consumers to deliberate over these consumption decisions. Voorhoeve (2013: 145) concludes that, since autonomy enhancing techniques are not effective in these circumstances, excise taxes have a soft paternalistic justification.

I disagree with this conclusion on the grounds that the automatic decisions of the intuitive system are not substantially non-voluntary. That intuitive system decisions do not involve effortful deliberation does not render them substantially non-voluntary. Voorhoeve argues that actions are substantially non-voluntary when agents are “insufficiently informed or inadequately capable of rational self- governance” (Voorhoeve 2013: 141). Projection bias caused by a lack of information does not have a soft paternalistic justification because it can be corrected with autonomy enhancing techniques. Thus, Voorhoeve is effectively arguing that the intuitive system is incapable of rational self-governance when exposed to current stimuli.

Price and Norman (2008: 29) argue that within the two systems framework (Epstein 1994; Kahneman 2002) there is a tendency to associate non-conscious processing with the intuitive system and conscious processing with the reasoning system. However, there is considerable ambiguity regarding how to map the notion of intuitive processing onto the distinction between conscious and non-conscious processes. They argue that the hallmark of consciousness is “control over the behavioural influence of information, and the ability to integrate the information in a flexible manner with changing contextual demands and executive goals” (2008: 30). Price and Norman’s use of conscious behaviour is extremely similar to autonomous behaviour in that both emphasise control and self-governance. This quality is supported by considerable empirical data (Jacoby et al. 1993; Merikle and Daneman 1998). The authors conclude that intuitive processes can be conscious processes. Taking Price and Norman’s definition of conscious processing, their argument implies that we have some form of control over our intuitive system. For example, intuitive decisions can often be shortcuts to conclusions we have repeatedly made through the reasoning system (Koriat 2000). If we exercise conscious control over intuitive decisions then we are capable of self-rule when making consumption choices regarding tobacco, alcohol and sugary drinks. This would imply that excise taxes do not have a soft paternalistic justification since our decisions are not substantially non- voluntary.

Shiffrin’s (1997) work also supports the conclusions of Price and Norman (2008). He argues that automatic processes can yield representational outcomes that we are consciously aware of. If, according to these authors, we can have some form of control over the intuitive system, then the decisions of this system are not substantially non-voluntary but are partially voluntary. Limiting these decisions undermines the goods of autonomy, albeit less than limiting the choices of the reasoning system. This view is supported by evidence that the reasoning system regulates the intuitive system (Kahneman 2002). If the intuitive system reaches a conclusion that is in conflict with our reasoning system then the reasoning system overrules that decision. This evidence gives weight to the notion that intuitive system decisions are somewhat autonomous and that we possess a capacity for self-governance even when using this system.

Voorhoeve (2013) and Rabin (2013) portray the intuitive and reasoning systems as dichotomous, separate entities. However, these systems should be seen as interdependent, mutually reinforcing partners. To argue that one system is fully conscious while the other is fully non-conscious does not consider the myriad of ways autonomy can be exercised. Intuitive feelings can be used to guide behaviour in a controlled and contextually sensitive manner even when biases are present (Price and Norman 2008). For example, imagine an individual who sees a child drowning in a river. This individual chooses to dive into the river to save the child. This decision was taken by the intuitive system as it was under extreme time pressure (Sunstein 2015). Luckily, they are able to save the child and return to land where they receive praise for their noble act (Wilkinson 2013). However, this individual also suffers from the overconfidence effect. This well-established bias causes a person’s subjective confidence in his or her ability to be greater than the objective assessment of that ability (Pallier et al. 2002). The overconfidence bias caused this individual to overstate the expected utility of saving the child because they underestimated the probability of drowning. The effect of overconfidence bias on this decision is the same as projection bias on consumption decisions, both cause an individual to exaggerate the expected utility of an action. I argue that this intuitive decision is partially voluntary and autonomous despite being subject to overconfidence bias. Most people would agree that the decision to save the child was autonomous and worthy of praise. If the decision taken by the intuitive system was substantially non-voluntary it would be inappropriate to praise this individual’s selfless courage. The individual’s intuitive decision reflects the values and beliefs held by their reasoning system (Sunstein 2015). This example shows that we regard intuitive system decisions to be autonomous and voluntary even if they are subject to bias. Analogously, intuitive system decisions which are subject to projection bias should be seen as partially voluntary.

This example rebuts Voorhoeve’s (2013) soft paternalistic justification, by showing that partially autonomous decisions can be made by the intuitive system even in the presence of bias. Consequently, excise taxes do not have a soft paternalistic justification since they inhibit partially voluntary action. Instead, excise taxes must be justified in a context wherein they diminish individual autonomy. To conclude, I have argued that justifications for excise taxes on tobacco, alcohol and sugary drinks cannot appeal to soft paternalism. Whilst I agree with Rabin’s model of how we acquire and sustain bad habits, I disagree that these decisions should be classified as substantially non-voluntary. Instead intuitive system decisions are partially voluntary since we can exercise some degree of conscious control over the intuitive system and regulate these decisions through the reasoning system. Thus, Rabin’s policies undermine the goods of autonomy and cannot appeal to soft paternalism.


  1. Arneson, R. J. 2005. Joel Feinberg and the justification of hard paternalism. Legal Theory 11(3): 259–284.
  2. Becker, G. S. and Murphy, K. M. 1988. A Theory of rational addiction. Journal of Political Economy 96(4): 675–700.
  3. Christman, J. 2012. Ben Colburn, Autonomy and Liberalism. Journal of Moral Philosophy 9(1): 134–136.
  4. Dworkin, G. 1988. The Theory and Practice of Autonomy. New York: Cambridge University Press.
  5. Epstein, S. 1994. Integration of the cognitive and psychodynamic unconscious. American Psychologist 49: 709–724.
  6. Feinberg, J. 1986. Harm to Self. Oxford: Oxford University Press.
  7. Jacoby, L. L., J. P. Toth and A. P. Yonelinas. 1993. Separating conscious and unconscious influences of memory: Measuring recollection. Journal of Experimental Psychology 122: 139–154.
  8. Kahneman, D. 2002. Maps of Bounded Rationality. Nobel Prize Lectures.
  9. Koriat, A. 2000. The feeling of knowing: Some metatheoretical implications for consciousness and control. Consciousness and Cognition 9: 149–171.
  10. Loewenstein, G., T. O’Donoghue and M. Rabin. 2003. Projection bias in predicting future utility. Quarterly Journal of Economics 118(4): 1209–48.
  11. Merikle, P. M. and M. Daneman. 1998. Psychological investigations of unconscious perception. Journal of Consciousness Studies 5: 5–18.
  12. Mill, J. S. 1991 [1859]. On Liberty and Other Writings, ed. S. Collini, Cambridge: Cambridge University Press.
  13. Pallier, G., Wilkinson, R., Danthiir, V., Kleitman, S., Knezevic, G., Stankov, L. and Roberts, R. 2002. The role of individual differences in the accuracy of confidence judgments. The Journal of General Psychology 129(3): 257--299.
  14. Price, M. C., and E. Norman. 2008. Intuitive decisions on the fringes of consciousness: Are they conscious and does it matter? Judgment and Decision Making 3(1): 28.
  15. Read, D. and B. Van Leeuwen. 1998. Predicting hunger: The effects of appetite and delay on choice. Organizational Behaviour and Human Decision Processes 76(2): 189–205.
  16. Rabin, M., 2013. Healthy habits: Some thoughts on the role of public policy in healthful eating and exercise under limited rationality. In Behavioural Public Policy, ed. A. Oliver, 115--139. Cambridge: Cambridge University Press.
  17. Shiffrin, R. M. 1997. Attention, automatism, and consciousness. In Scientific Approaches to Consciousness, eds. J. D. Cohen and J. W. Schooler, 49–64. New Jersey, NJ: Lawrence Erlbaum Associates.
  18. Sunstein, C. R. 2015. The Ethics of Nudging. Yale Journal on Regulation 32(2): 413—450.
  19. Voorhoeve, A. 2013. Response to Rabin. In Behavioural Public Policy, ed. A. Oliver, 140--147. Cambridge: Cambridge University Press.
  20. Wilkinson, T. M. 2013. Nudging and manipulation. Political Studies 61(2): 341--355.