Academic journal article The Psychological Record

Categorical Discrimination of Sequential Stimuli: All [S.Sup.[DELTA]] Are Not Created Equal

Academic journal article The Psychological Record

Categorical Discrimination of Sequential Stimuli: All [S.Sup.[DELTA]] Are Not Created Equal

Article excerpt

Categorical coherence, the extent to which exemplars from the same category resemble one another more than they resemble exemplars from other categories (Astley & Wasserman, 1992), has been shown to promote faster category learning in nonhumans. In laboratory discrimination tasks, pigeons (Astley & Wasserman, 1992; Lazareva, Soto, & Wasserman, 2010) and other animals (Roberts & Mazmanian, 1988) made more errors when negative stimuli resembled positive stimuli than when they did not. For example, if positive stimuli were images of flowers (such that a response following the presentation of the image produced food), rhesus monkeys were more likely to respond to images of trees than to images of faces (Sands, Lincoln, & Wright, 1982).

In slot machine play and other types of gambling, near wins are outcomes in which the stimuli presented are visually similar to wins, but there is no payout. For example, if five cherries are a win that pays out in a slot machine, four cherries and a lemon and no payout would be a near win. Wins and near wins have low categorical coherence because they look similar even though they are not functionally equivalent, whereas wins and other stimulus arrangements that signal losses have high categorical coherence. Near wins have been shown to facilitate continued betting in the absence of wins (e.g., Cote, Caron, Aubert, Desrochers, & Ladouceur, 2003; Strickland & Grote, 1967). Their presence has been associated with greater persistence betting in computerized slot machines (Harrigan, 2007; Kassinove & Schare, 2001) and human gamblers have described them as more similar to wins than to losses (Dixon & Schreiber, 2004). Such near-win effects can be considered failures of categorical coherence.

The mechanism driving near-win effects has been long-sought in behavioral research on gambling because of its potential importance to the development of problem gambling (Strickland & Grote, 1967). Like other failures of categorical coherence, near-win effects are often interpreted as problems of stimulus generalization (Belisle & Dixon, 2016) or conditioned reinforcement (Skinner, 1953; Witts, Ghezzi, & Manson, 2015). Disambiguating these possibilities is a challenge: stimulus discriminability is a factor in the determination of conditioned reinforcing value (Davison & Nevin, 1999), so separate measures of stimulus control and value can be difficult to obtain.

Due to its potential translational value, animal behavior researchers have turned their attention to the description of the structural characteristics of near-win effects in recent years. Laboratory animal models have several potential advantages over studies of more naturalistic human gambling behavior (Madden, Ewan, & Lagorio, 2007). Laboratory animals can be studied for extended periods of time in highly controlled experimental conditions. Although operant-conditioning preparations can lack ecological and face validity, they allow investigators to isolate potential controlling variables experimentally and establish causal links between specific stimuli or environmental contexts and behavioral outcomes.

Slot machine analog experiments with rats typically involve a go/no-go task with one positive stimulus and multiple negative stimuli that differ from the positive on a single, quantifiable dimension (e.g., number of activated lights). The rat presses a lever to begin a trial, an action that is loosely analogous to betting. On winning trials, the rat must press another lever to collect the food payout. In one rat model of slot machine gambling (Peters, Hunt, & Harper, 2010), one to five chamber lights were lit as a consequence of the rat pressing a lever designed to simulate placing a bet in a slot machine or video lottery terminal. Trials with five lights presented at the beginning of a trial signaled a large win and four lights signaled a small win. All other trials (i. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.