THE ATTRACTION OF GAMBLING HOWARD RACHLIN,VASILIY SAFIN,KODI B. ARFER, AND MING YEN STONY BROOK UNIVERSITY If a repeated gamble is subjectively structured into units each consisting of a string of consecutive losses followed by a single win, longer strings will necessarily be less valuable. Longer, less valuable strings will be discounted by delay more than will shorter, more valuable strings. This implies that the whole gamble’s expected, delay-discounted value will increase as delay discounting increases. With this restructuring, even games of (objectively) negative expected value, such as those at casinos, may be subjectively positive. The steeper the delay discounting, the greater the subjective value of the gamble (over normal ranges of discounting steepness). Frequent gamblers, who value gambles highly, would thus be expected to discount delayed rewards more steeply than would nongamblers. Key words: addiction, delay discounting, gambling, patterning of behavior, probability discounting, restructuring, variable-ratio schedules Even strictly economic models of addiction (e.g., Becker & Murphy, 1988) presume the existence of a commodity that addicts consume at an unusually high rate. An alcohol addict consumes alcohol; if wine is preferred but unavailable, whiskey or beer will do. If a person has an unusually strong preference for immedi- ate over delayed consumption of some com- modity, the person would be unusually susceptible to addiction to that commodity. But in the case of gambling there is no immediately consumable commodity to be addicted to. Winning money is not like drinking alcohol; alcohol is a primary reinforcer while winning money depends on potential exchange for its value. While a pathological gambler may also be a “big spender,” spending in no way substitutes for gambling; it cannot be said that gamblers consume money in the way alcoholics consume alcohol. Pathological gamblers do seem to seek risk, but risk is not a commodity so much as a pattern of interaction between behavior and environment. Nevertheless patho- logical gambling has many properties of an addiction. Just as heroin, alcohol, and smoking addicts discount delayed rewards more steeply than do nonaddicts (Odum, 2011), so patholog- ical gamblers discount delayed rewards more steeply than do nongamblers (Alessi & Petry, 2003; Dixon, Marley, & Jacobs, 2003; Petry, 2001; Petry & Madden, 2010; Reynolds, 2006). It is not clear why this should be the case. When there is a delay between a bet and its outcome (as at a racetrack) one might think that a person with a steep delay discount function would prefer the lesser amount of money that they have (zero delay) to the larger but delayed amount that they might win. How then might an abnormally high valuation of immediate over delayed rewards result in an abnormally high preference for gambles over sure things? This article’s purpose is to answer that question. Just as delay decreases reward value, so does uncertainty. The choice between certain and probabilistic rewards parallels the choice between immediate and delayed rewards. However, behavioral as well as cognitive studies of individual choice between small-certain rewards and larger probabilistic rewards (e.g., Kahneman & Tversky, 1979; Rachlin, Raineri & Cross, 1991) often find risk avoidance (i.e., discounting of probabilistic rewards below their expected value). They generally do not find the risk-seeking that would explain why a person would choose a gamble over a nongamble. Myerson, Green, Hanson, Holt, and Estle (2003) did find that college students who said they gambled discounted probabilistic rewards significantly less steeply than students who This research was supported by grant DA02652021 from the National Institute of Drug Abuse. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This article is a revision and expansion of Rachlin, 1990. We thank David Cross for many helpful suggestions and criticisms. Address correspondence to Howard Rachlin, Psychology Department, Stony Brook University, Stony Brook, NY 11794. E-mail: howard.rachlin@stonybrook.edu. doi: 10.1002/jeab.113 JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 2014, 9999, 1–7 NUMBER 9999 (OCTOBER) 1