Pepper & Nettle's (P&N's) article is an important contribution to a growing body of research that shows that poor people frequently make very sensible choices, given the enormous constraints that they generally face (Banerjee & Duflo Reference Banerjee and Duflo2011; Collins et al. Reference Collins, Morduch, Rutherford and Ruthven2010; Mullainathan & Shafir Reference Mullainathan and Shafir2013; Nettle Reference Nettle2010a). It is exciting to see further integration of life-history theory with the study of poverty and the conditions of deprivation (Jones Reference Jones2015), especially coming from some of the leaders of the study of life-history decisions among contemporary people (Nettle Reference Nettle2010a).

P&N do not specifically discuss uncertainty, but it is central to the lack of control that lies at the heart of the behavioral constellation of deprivation (BCD) and is, in all likelihood, the force that shaped the evolution of the constellation itself (Jones Reference Jones2011; Reference Jones2015). In the decision-theoretic sense, “risk” is defined as a variable reward, as might arise through a lottery. Risk implies known probabilities of the potential payoffs. “Uncertainty,” on the other hand, implies that the probabilities of the variable rewards are not known. Indeed, the very outcome of the decision might not be known (Knight Reference Knight1921).

Temporal discounting is a way of accounting for time preferences. P&N present ample evidence for a preference for immediate rewards in the BCD. But two key features of discounting need to be explained by any theory of time preference: the overall magnitude of temporal discounting and the shape of the discount function. The contextually appropriate response (CAR) hypothesis makes a clear prediction about the magnitude of discounting, but it is less clear about the shape of the discount function. This is an area where a focus on uncertainty helps fill a theoretical gap. Higher discount rates will lead to a strong preference for immediate rewards, consistent with CAR. The shape of the discount function – how discounting changes with time – is a distinct phenomenon. The orthodox theory of economic choice, based on the maximization of expected utility, requires a constant discount rate. This constant rate in continuous time leads to an exponential discount function, and this particular form is crucial for expected utility theory, because only an exponential discount function will lead to dynamic consistency (i.e., in the absence of new information, preference orderings at the outset will remain consistent as time proceeds). A substantial empirical literature suggests that people frequently violate the expectation of constant time preference (Ainslie Reference Ainslie1975; Herrnstein Reference Herrnstein1961; Laibson Reference Laibson1997). People are typically anchored in the present: Any delay from the immediate is discounted steeply. However, as time passes, they show decreasing impatience. The apparent discount rate declines over time as people become indifferent to small delays in the more remote future. This pattern of discounting is typically referred to as “hyperbolic,” although a more descriptive term is “present bias/diminishing impatience” (PB/DI).

Starting with Ainslie (Reference Ainslie1975), PB/DI was generally interpreted as a failure of willpower, because small rewards with short waiting times can be favored over larger rewards with longer waiting times, even if the larger rewards are initially favored. Many of the sensation-seeking features of the BCD have been linked to this putatively irrational, time inconsistency (Bernheim et al. Reference Bernheim, Ray and Yeltekin2015; Loewenstein Reference Loewenstein1996). However, recent work incorporating uncertainty into the theory of time preferences suggests that PB/DI can arise in rational agents when the future is uncertain and information available for learning about the future is limited (or very costly to acquire). Yaari (Reference Yaari1965) notes that mortality hazard could mimic the effect of a pure time preference. Sozou (Reference Sozou1998) notes that the constant time preference of expected utility theory is formally equivalent to an agent's having a prior distribution for failure times with an exponential hazard. He further shows that integrating across uncertainty in this otherwise time-consistent framework naturally leads to PB/DI. The clear conclusion from a number of authors, using a variety of formal models, is that uncertainty leads to PB/DI, even if the underlying intertemporal preferences are consistent (Dasgupta & Maskin Reference Dasgupta and Maskin2005; Halevy Reference Halevy2008; Jackson & Yariv Reference Jackson and Yariv2014; Sozou Reference Sozou1998).

There is a surprising lack of empirical evidence on the association between deprivation and differences in the shape of discount functions. However, the association between PB/DI and uncertainty potentially provides an important avenue for future empirical and experimental work. People with greater uncertainty should be more present-biased in addition to having higher discount rates overall.

I have suggested that the unusual features of the human life cycle (late age at first reproduction [AFR] coupled with high fertility, overlapping periods of dependence, extensive alloparental care, and postreproductive survival) are consistent with adaptations to extreme environmental variability (Jones Reference Jones2005; Reference Jones2011). In brief, stochastic demography (Tuljapurkar Reference Tuljapurkar1990) shows that variance in vital rates induced by environmental variation is generally bad for fitness, and, when the variance primarily affects juvenile survival, the negative effects are modulated by generation time. Longer generation times (which arise from late AFR and long reproductive spans) smooth over environmental variation, reducing the negative effects. Selection is therefore expected to lengthen the life cycle and lead to bet-hedging reproductive tactics (Jones Reference Jones2011). The origin of the genus Homo and the evolution of modern H. sapiens are associated with a broad cooling and drying of the planet and, especially, periods of substantial environmental instability that have, at times, been extremely high frequency (Anklin et al. Reference Anklin, Barnola, Beer, Blunier, Chappellaz, Clausen, Dahljensen, Dansgaard, Deangelis, Delmas, Duval, Fratta, Fuchs, Fuhrer, Gundestrup, Hammer, Iversen, Johnsen, Jouzel, Kipfstuhl, Legrand, Lorius, Maggi, Miller, Moore, Oeschger, Orombelli, Peel, Raisbeck, Raynaud, Schotthvidberg, Schwander, Shoji, Souchez, Stauffer, Steffensen, Stievenard, Sveinbjornsdottir, Thorsteinsson and Wolff1993; deMenocal Reference deMenocal1995; Reference deMenocal2004). Humans are adapted to a cool and unpredictable planet; as such, we should expect substantial adaptations to managing uncertainty. This highlights the potential importance of the BCD for understanding human life histories in general – not simply in industrial nation-states – and the central role that uncertainty has played in shaping these adaptive responses.