Reinforcement Frequency and Missed-Opportunity Framing in Repeated Risky Choice

Abstract: Under standard economic theory, incentives that imply the same distribution of final payoffs should affect choice in the same way. In repeated choice, however, the same chance of winning can be delivered through frequent lottery tickets, each with a small chance of winning, or through rare tickets, each with a larger chance. This thesis asks whether such economically equivalent lotteries affect repeated risky choice differently, and whether, once a ticket has been missed, the way the missed ticket is described influences what people choose next. Two preregistered online studies used a repeated-choice task of 100 trials. Study 1 (N = 163) directly replicated Study 1 of Luria, Erev, and Haruvy (2017). Participants repeatedly chose between a safe option and a risky option with the same expected value in all conditions. In a control condition, the risky option paid the prize directly; in three lottery conditions, it produced a lottery ticket on 1/20, 1/3, or 2/3 of risky choices, and the chance that each ticket won fell accordingly. Risky choice increased with ticket frequency, from 3% of trials in the control condition to 46% in the 2/3 condition (η² = .37). In the three lower-frequency conditions, participants gradually abandoned the risky option, consistent with the hot-stove effect, whereas frequent tickets kept it in use throughout the task. Participants in the 2/3 condition also differed markedly: some chose the risky option almost always and others almost never. Study 2 (N = 243) kept the three lottery conditions and added feedback after each safe choice on which the risky option would have produced a ticket. The feedback was either a factual reminder that the other option could produce a ticket or a regret-enhancing message stating that a ticket had just been missed. The frequency effect replicated (η² = .21). Participants who received the regret-enhancing message chose the risky option more often than those who received the factual reminder (45% vs. 36% of trials, d = 0.31). In an exploratory analysis, this difference was absent when tickets were rare and largest when they were frequent. Unlike in Study 1, risky choice did not decline over the task. Self-reported regret, however, did not differ between the two messages. Together, the findings show that the effect of a lottery incentive on repeated choice depends not only on its expected value but also on how often it produces intermediate rewards, and that the wording of feedback about a missed opportunity can change later choice even when it reports the same missed ticket. Because the regret-enhancing message did not increase reported regret, the mechanism behind this framing effect remains open