|
| Rational and Irrational Decision Making: Is the Subject Always Right? |
| Tuesday, June 1, 2010 |
| 9:00 AM–10:20 AM |
| Republic A (Grand Hyatt) |
| Area: EAB/TPC; Domain: Experimental Analysis |
| Chair: Jennifer Rusak (University of Florida) |
| Abstract: Research in human decision making has identified systematic deviations from rational choice predictions. The current symposium will present experiments relating to four domains in this research: the sunk cost fallacy, framing and loss aversion, the prisoner's dilemma, and hyperbolic discounting of delayed outcomes. Macaskill will discuss the conditions under which pigeons and humans make choices based on past investment rather than future outcomes. Rusak will present data indicating the extent to which pigeons' choices between gains and losses are influenced by the framing of alternatives. Locey will show that pigeons learned to cooperate in an iterated prisoner's dilemma game despite higher local rates of reinforcement for defection. Myerson will demonstrate that hyberboloid functions obtained from pigeons in a probability discounting procedure were comparable to those obtained using delay discounting procedures. Each of these papers contributes to the broader literature on decision making by extending the methods used to study it and describing conditions under which relevant phenomena are observed. In addition, cross-species analyses have the potential to clarify whether deviations from rational choice predictions result from fundamental behavioral processes, or whether they are unique to humans (and perhaps reflect verbal mediation). |
| |
| Contingency Sensitivity and the Sunk-Cost Fallacy in Humans and Pigeons |
| ANNE C. MACASKILL (University of Florida), Timothy D. Hackenberg (Reed College) |
| Abstract: The sunk cost effect occurs when non-optimal persistence is controlled by past investments. The current study investigated this fallacy, and related decisions about persistence, using a laboratory model with pigeons and humans. On each trial, one of four (unsignaled) ratios was in effect; subjects could either continue working on the current ratio, or make a reset response and draw a new ratio. Ratio sizes, the probability at which they occurred, and therefore the relative pay-offs of persisting and resetting varied across conditions. This allowed an assessment of the features of this contingency arrangement that controlled behavior. Both pigeons and humans behaved most optimally (in terms of unit price) when the contingencies strongly favored either persisting or resetting; however when payoffs only slightly favored resetting they often committed the sunk cost error. Humans were more likely to make the reverse error and reset when persisting was optimal. |
| |
| An Evaluation of Framing Effects and Loss Aversion in Pigeons |
| JENNIFER RUSAK (University of Florida), Timothy D. Hackenberg (Reed College) |
| Abstract: Research in human decision-making has identified systematic deviations from rational choice predictions. For example, it has been shown that individuals tend to avoid losses, even when alternatives differ only in whether they are framed as a gain or loss. Little is known about the generality of framing effects and loss aversion with non-human animals. The purpose of the present investigation was to parametrically examine framing effects and loss aversion. Four pigeons made repeated choices between gain or loss alternatives in which tokens were probabilistically added or subtracted from a starting amount such that average payoffs for each alternative either differed or were equated across condition. In another phase, choices were made between an alternative with a fixed number of tokens or one in which some number of tokens was subtracted with certainty. Data indicate that for all subjects, choices were sensitive to token amount manipulations, but less systematic results were observed in critical conditions in which average payoffs were equivalent across alternative. The results of this investigation provide preliminary data on the generality of framing effects and loss aversion seen with human subjects to non-human subjects, and illustrate promising methods for examining framing and loss aversion in the animal laboratory. |
| |
| Pigeons as Prisoners in a Prisoner’s Dilemma Game: A Molar Versus Molecular Analysis of Choice |
| MATT LOCEY (Stony Brook University), Howard Rachlin (Stony Brook University) |
| Abstract: Research in various domains of the experimental analysis of behavior has addressed the issue of whether behavior is controlled by immediate or global consequences. The prisoner’s dilemma game, a classic problem in game theory, can been used to address this same issue. In a prisoner’s dilemma, two subjects are paired together and given the opportunity to “cooperate” or “defect”. The immediate consequences of each choice always favor defecting over cooperating (at least, for the individual that is defecting). However, if both players defect, the consequences are worse (for both) than if both cooperated. In the present experiment, four pigeons were paired together to participate in an iterated prisoner’s dilemma game. Each pigeon chose between cooperating on an FR 23 and defecting on an FR 15. Each cooperation response reduced the ratio requirements for the other pigeon. Under these conditions, all 4 pigeons eventually learned to cooperate despite the higher local rate of reinforcement for defection. |
| |
| Pigeons’ Discounting of Probabilistic and Delayed Reinforcers |
| Amanda L. Calvert (Washington University in St. Louis), Leonard Green (Washington University), JOEL MYERSON (Washington University) |
| Abstract: Pigeons’ discounting of probabilistic and delayed food rewards was studied using adjusting-amount procedures. In the probability discounting conditions, pigeons chose between an adjusting number of food pellets contingent on a single key peck and a larger, fixed number of pellets contingent on completion of variable-ratio schedules, which represent repeated gambles. In the delay discounting conditions, pigeons chose between an adjusting number of pellets delivered immediately and a larger, fixed number of pellets delivered after a delay. The major finding of the present study is that probability discounting (as a function of the odds against reinforcement) was typically as well described by a hyperboloid function as was delay discounting (as a function of the time until reinforcement), although the exponents for fits to the probability discounting data were lower than the exponents for fits to the delay discounting data, a result similar to that observed with humans. For five of the eight pigeons, the hyperboloid function that best fit the data from the delay discounting task also described the subjective values of probabilistic reinforcers as a function of the geometric means of the delays to those reinforcers. We consider the implication of these findings for Rachlin, Logue, Gibbon, and Frankel’s (1986) hypothesis that the discounting of probabilistic rewards on repeated gambles is controlled by the average time until a win. |
|
| |