Sequential Decisions from Sampling: Inductive Generation of Stopping Decisions Using Instance-Based Learning Theory

Sequential decisions from sampling are common in daily life: we often explore alternatives sequentially, decide when to stop such exploration process, and use the experience acquired during sampling to make a choice for what is expected to be the best option. In decisions from experience, theories of sampling and experiential choice are unable to explain the decision of when to stop the sequential exploration of alternatives. In this chapter, we propose a mechanism to inductively generate stopping decisions, and we demonstrate its plausibility in a large and diverse human data set of the binary choice sampling paradigm. Our proposed stopping mechanism relies on the choice process of a theory of experiential choice, Instance-Based Learning Theory (IBLT). The new stopping mechanism tracks the relative prediction errors of the two options during sampling, and stops when such difference is close to zero. Our results from simulation are able to accurately predict human stopping decisions distributions in the dataset. This model provides an integrated theoretical account of decisions from experience, where the stopping decisions are generated inductively from the sampling process.

What are the chances? How descriptions of probability affect risky decisions

<p>People make different decisions when they know the odds of an event occurring, (e.g. told 10% chance of an earthquake that year) than when they draw on only their own experience (e.g. living in a city with, on average, one earthquake every 10 years). It may be that when we make decisions based on our past experience (decisions from experience) we are more likely to choose a risky option when it can lead to the biggest win and avoid it when it can lead to the biggest loss, this effect is called the Extreme-Outcome rule. Across three Experiments we tested the Extreme-Outcome rule by having participants make repeated choices between either safe or risky options which had the same expected value. In each experiment, we varied the magnitude of the reinforcer’s participants could win in both an Experience condition and a condition that had both description and experience information. In Experiment 1 where we had two reinforcer sizes (small and large) we found an Extreme-Outcome effect in the Experience condition, but not the Description-Experience condition. In Experiment 2 we tested a prediction of the Extreme-Outcome rule that participants would be sensitive to the best and worst outcome by adding another reinforcer size (reinforcers were small, medium and large) and therefore on some trials neither alternative included an extreme outcome. We also removed zero as a potential outcome to investigate whether zero aversion might be driving the effect of reinforcer magnitude in the Experience condition. We did not find response patterns consistent with an Extreme-Outcome rule in the Experience condition. Instead, participants were least risk seeking when the reinforcer was small, but there was no difference in levels of risk seeking between the medium and large reinforcer trials. In other words, there was an effect of the low-extreme outcome but not the high-extreme outcome. Like Experiment 1, in the Description-Experience condition risk preference was not influenced by reinforcer size, but the absolute levels were higher. To investigate whether this increase in risk preference was due to removing the zero, in Experiment 3 we manipulated whether zero was present or absent. When zero was absent, risk preference was not influenced by the size of the reinforcer in the Description-Experience condition, but there was an effect of the low-extreme outcome when zero was present. We also found an effect of the low extreme outcome in the Experience condition regardless of whether zero was present or absent. Overall, these findings suggest the Extreme-Outcome rule needs to be modified to take into account the effect of the low extreme but not the high extreme outcome.</p>

What are the chances? How descriptions of probability affect risky decisions

<p>People make different decisions when they know the odds of an event occurring, (e.g. told 10% chance of an earthquake that year) than when they draw on only their own experience (e.g. living in a city with, on average, one earthquake every 10 years). It may be that when we make decisions based on our past experience (decisions from experience) we are more likely to choose a risky option when it can lead to the biggest win and avoid it when it can lead to the biggest loss, this effect is called the Extreme-Outcome rule. Across three Experiments we tested the Extreme-Outcome rule by having participants make repeated choices between either safe or risky options which had the same expected value. In each experiment, we varied the magnitude of the reinforcer’s participants could win in both an Experience condition and a condition that had both description and experience information. In Experiment 1 where we had two reinforcer sizes (small and large) we found an Extreme-Outcome effect in the Experience condition, but not the Description-Experience condition. In Experiment 2 we tested a prediction of the Extreme-Outcome rule that participants would be sensitive to the best and worst outcome by adding another reinforcer size (reinforcers were small, medium and large) and therefore on some trials neither alternative included an extreme outcome. We also removed zero as a potential outcome to investigate whether zero aversion might be driving the effect of reinforcer magnitude in the Experience condition. We did not find response patterns consistent with an Extreme-Outcome rule in the Experience condition. Instead, participants were least risk seeking when the reinforcer was small, but there was no difference in levels of risk seeking between the medium and large reinforcer trials. In other words, there was an effect of the low-extreme outcome but not the high-extreme outcome. Like Experiment 1, in the Description-Experience condition risk preference was not influenced by reinforcer size, but the absolute levels were higher. To investigate whether this increase in risk preference was due to removing the zero, in Experiment 3 we manipulated whether zero was present or absent. When zero was absent, risk preference was not influenced by the size of the reinforcer in the Description-Experience condition, but there was an effect of the low-extreme outcome when zero was present. We also found an effect of the low extreme outcome in the Experience condition regardless of whether zero was present or absent. Overall, these findings suggest the Extreme-Outcome rule needs to be modified to take into account the effect of the low extreme but not the high extreme outcome.</p>

Frequency of enforcement is more important than the severity of punishment in reducing violation behaviors

External enforcement policies aimed to reduce violations differ on two key components: the probability of inspection and the severity of the punishment. Different lines of research offer different insights regarding the relative importance of each component. In four studies, students and Prolific crowdsourcing participants (Ntotal = 816) repeatedly faced temptations to commit violations under two enforcement policies. Controlling for expected value, we found that a policy combining a high probability of inspection with a low severity of fines (HILS) was more effective than an economically equivalent policy that combined a low probability of inspection with a high severity of fines (LIHS). The advantage of prioritizing inspection frequency over punishment severity (HILS over LIHS) was greater for participants who, in the absence of enforcement, started out with a higher violation rate. Consistent with studies of decisions from experience, frequent enforcement with small fines was more effective than rare severe fines even when we announced the severity of the fine in advance to boost deterrence. In addition, in line with the phenomenon of underweighting of rare events, the effect was stronger when the probability of inspection was rarer (as in most real-life inspection probabilities) and was eliminated under moderate inspection probabilities. We thus recommend that policymakers looking to effectively reduce recurring violations among noncriminal populations should consider increasing inspection rates rather than punishment severity.

The Effect of Time Pressure on Risky Financial Decisions from Description and Decisions from Experience

Time pressure has been found to impact decision making in various ways, but studies on the effects time pressure in risky financial gambles have been largely limited to description-based decision tasks and to the gain domain. We present two experiments that investigated the effect of time pressure on decisions from description and decisions from experience, across both gain and loss domains. In description-based choice, time pressure decreased risk seeking for losses, whereas for gains there was a trend in the opposite direction. In experience-based choice, no impact of time pressure was observed on risk-taking, suggesting that time constraints may not alter attitudes towards risk when outcomes are learned through experience.

The Effect of Time Pressure on Risky Financial Decisions from Description and Decisions from Experience

Time pressure has been found to impact decision making in various ways, but studies on the effects time pressure in risky financial gambles have been largely limited to description-based decision tasks and to the gain domain. We present two experiments that investigated the effect of time pressure on decisions from description and decisions from experience, across both gain and loss domains. In description-based choice, time pressure decreased risk seeking for losses, whereas for gains there was a trend in the opposite direction. In experience-based choice, no impact of time pressure was observed on risk-taking, suggesting that time constraints may not alter attitudes towards risk when outcomes are learned through experience.

Prior Beliefs and Ambiguity Attitudes in Decision from Experience

Prior beliefs and their updating play a crucial role in decisions under uncertainty, and theories about them have been well established in classical Bayesianism. Yet, they are almost absent for ambiguous decisions from experience. This paper proposes a new decision model that incorporates the role of prior beliefs, beyond the role of ambiguity attitudes, into the analysis of such decisions. Hence, it connects ambiguity theories, popular in economics, with decision from experience, popular (mostly) in psychology, to the benefit of both. A reanalysis of some existing data sets from the literature on decisions from experience shows that the model that incorporates prior beliefs into the estimation of subjective probabilities outperforms the commonly used model that approximates subjective probabilities with observed relative frequencies. Controlling for subjective priors, we obtain more accurate measurements of ambiguity attitudes, and thus a new explanation of the gap between decision from description and decision from experience. This paper was accepted by Manel Baucells, decision analysis.