The sign effect, systematic devaluations and zero discounting

Abstract

The sign effect is the steeper discounting of gains compared to losses. However, we see a greater sign effect on an individual level compared to an aggregate level. In this experiment, we compare discounting of gains and losses on an individual and aggregate level, to explore further details about when and to what extent human adults discount. Thirty-one participants went through a computer-based choice-task procedure of hypothetical monetary gains and losses. The results show clear qualitative differences between discounting of gains and losses, adding empirical data to support the sign effect. The results also support previous findings that show that aggregate and individual results do not always correspond. Further, the within-subject details showing zero discounting or nonsystematic changes concerning losses, replicate earlier studies suggesting that discounting of gains and losses involve different reinforcing contingencies. The present study expands on this research area by including verbal reports, supplementing details about the unobserved reinforcing contingencies. Implications of research on discounting may indicate how to deal with decision-making challenges and might shed light on why predictions about complex decision making sometimes fail.

Sign effect in adolescents: Within‐subject comparison of delay discounting of hypothetical monetary gains and losses

“…The aggregate discounting curves and their corresponding area-under-the-curve measures in this experiment support the finding that discounting of gains appears steeper than discounting of losses also among adolescents. The differences are even more profound among these adolescents compared to previously obtained results from adults (Furrebøe, 2020). However, there are features in the results contradicting the discounting paradigm, and consequently the sign effect.…”

“…In order to be able to obtain a discounting pattern, we used a choice‐based technique consisting of binary comparisons between Smaller–Sooner (SS) and Larger–Later (LL) hypothetical alternatives. The current procedure was based on a previously used “double‐delay” procedure (Holt et al, 2008), and adjusted to fit discounting tasks for both gains and losses (Furrebøe, 2020).…”

“…Recent studies have also shown that the willingness to wait for a gain was different between adolescents with and without ADHD, but the willingness to wait for a (smaller) loss was not different between the groups (Mies et al, 2019). Further, studies report that while some adult participants discount losses similar to how they discount gains, others may show “negative” discounting (Hardesty et al, 2013), increasingly/decreasingly preferring sooner rather than later losses, or “zero” discounting of losses (Furrebøe, 2020), choosing the sooner loss regardless of delay. These results are consistent with other studies demonstrating qualitative differences between discounting of gains and losses.…”

“…A procedure where both outcomes are delayed and adjusted, emphasizes these relative values, yet recognizes the effect of immediacy. Only a few discounting studies have used a procedure with two delayed gains (Green et al, 2005; Kable & Glimcher, 2010; Mitchell & Wilson, 2010), two delayed losses (Holt et al, 2008), or two delayed gains and losses (Furrebøe, 2020). Green et al (2005) were the first to show that discounting obtained when choices involve two delayed rewards can also be described by the hyperbolic function.…”

“…Studies examining whether the qualitative and quantitative differences previously found in discounting of gains and losses (Myerson et al, 2017), using a “double‐delay” procedure is called for. Specifically, it remains to be seen if such a procedure conducted on adolescents will yield similar discounting patterns as for adults (Furrebøe, 2020). If the sign effect, in fact, is a display of different competing contingencies involving the effects of immediacy of the outcomes as well as considerations of relative values of delayed outcomes, we expect at least as large a sign effect on an aggregate level, and equally large qualitative variations on an individual level, to appear in adolescents as we have seen in adults.…”