Reinforcement Contingencies and Signal Detection

A chamber containing 72 response keys defining the circumference of a circle 1 m in diameter was used to examine the relation between differentiation of response location and a measure of responsereinforcer contingency known as the phi coefficient. A different target key was specified in each successive phase, and response location was differentiated with respect to the target. Criterional and noncriterional responses (i.e., responses "near" and "far" from the target) were defined using targeted percentile schedules to control the overall probability of each response class. By manipulating criterional (and, hence, noncriterional) response probability and the reinforcement probabilities conditional on each, a mathematical invariance property peculiar to phi in contingency analysis was examined. Specifically, diagonally interchanging cell frequencies in a 2 x 2 table relating criterional/noncriterional responses to reinforcement/nonreinforcement leaves phi unchanged. Hence, the degree of response differentiation predicted by phi remains unchanged under the four permutations implied by the various diagonal interchanges. This predicted invariance was examined under values of phi equal to .33, .58, and .82. Increasing phi generally increased the stereotypy of response location. Three of the permutations generated almost interchangeable performance at different phi values. The remaining permutation, however, generated functions relating response concentration to phi with slopes shallower than those obtained under the other permutations. This resulted from relatively higher levels of differentiation, compared to the other permutations, at low phi values. These data strongly suggest boundary conditions on the ability of phi to reflect completely the local processes that are indexed by phi at a molar level.Key words: response-reinforcer contingency, response differentiation, shaping, differential reinforcement, percentile schedules, response location, pigeonsThe term contingency suffers from use in multiple contexts in the analysis of behavior, most particularly in operant conditioning. In a general sense, contingency refers to some relation between a response and a consequence, as in the phrases response-contingent or contingencies of reinforcement. The latter often substitutes for schedule of reinforcement, and as such, changes in schedule parameters need not change the contingency as long as the type of schedule remains the same. The statistical meaning of contingency, alternatively,

Section: Resultsmentioning

confidence: 99%

Response—reinforcer Contingency and Spatially Defined Operants: Testing an Invariance Property of Phi

Galbicka

Platt

1989

“…Davison and Tustin (1978) arrived at essentially the same bias-free measure of sample-stimulus discriminability as that of classical signal-detection theory (dЈ, Green & Swets, 1966) by adapting an empirically-validated quantitative description of simple choice known as the generalized matching law (GML, Baum, 1974, 1979. They argued that choice between responding ''yes'' or ''no'' in the presence and absence of the signal would be biased toward either alternative to a degree that depended on the discriminability of the signal from noise and the history of reinforcement for each response (see Nevin, Jenkins, Whittaker & Yarensky, 1982, for the same argument). The GML was used to describe the effects of both the frequency with which responses had been reinforced and inherent biases, and another parameter was added to account for the bias that was attributable to discriminating the signal from noise (or sample stimuli).…”

mentioning

confidence: 99%

Quantitative Analyses of Matching‐to‐sample Performance

2003

Six pigeons performed a simultaneous matching-to-sample (MTS) task involving patterns of dots on a liquid-crystal display. Two samples and two comparisons differed in terms of the density of pixels visible through pecking keys mounted in front of the display. Selections of Comparison 1 after Sample 1, and of Comparison 2 after Sample 2, produced intermittent access to food, and errors always produced a time-out. The disparity between the samples and between the comparisons varied across sets of conditions. The ratio of food deliveries for the two correct responses varied over a wide range within each set of conditions, and one condition arranged extinction for correct responses following Sample 1. The quantitative models proposed by Davison and Tustin (1978), Alsop (1991), and failed to predict performance in some extreme reinforcer-ratio conditions because comparison choice approached indifference (and strong position biases emerged) when the sample clearly signaled a low (or zero) rate of reinforcement. An alternative conceptualization of the reinforcement contingencies operating in MTS tasks is advanced and was supported by further analyses of the data. This model relates the differential responding between the comparisons following each sample to the differential reinforcement for correct responses following that sample.

“…The present experiment also provides further information relevant to recent attempts to integrate the matching-law and signal-detection analyses of choice in one quantitative choice model (e.g., McCarthy & Davison, 1981;Nevin et al, 1982). First, free-operant procedures can be used to examine these models.…”

Section: Resultsmentioning

confidence: 83%

“…One approach has involved the construction of mathematical models incorporating elements of both paradigms (e.g., McCarthy & Davison, 1981;Nevin et al, 1982 (see, e.g., Elsmore, 1972;Killeen, 1978;Stubbs, 1976a). Yet all of these experiments allowed only one response per trial and none of them scheduled reinforcers according to variable-interval schedules, standard matching-law procedure.…”

Section: Rationalementioning

confidence: 99%

Signal Detection and Matching: Analyzing Choice on Concurrent Variable‐interval Schedules

Logue

1983

Pigeons' pecks on a red key and a green key were followed by access to grain according to pairs of concurrent independent variable-interval schedules in a combined signal detection/ matching law paradigm. Pecks on the red key were reinforced by the richer variable-interval schedule if a short-duration tone had been presented; pecks on the green key were reinforced by the richer variable-interval schedule if a long-duration tone had been presented. Pecks on the green key given a short-duration tone, or on the red key given a long-duration tone, were reinforced by the leaner variable-interval schedule. The data were analyzed according to both signal detection's and the matching law's separate measures of, first, the discrimination of the choices and, second, the bias to make one response or another. Increasing the difficulty of the tone-duration discrimination decreased both methods' measures of the discrimination of the choices and did not change both methods' measures of the bias to make one response or another. Changing the leaner variable-interval schedule so that it approached the richer variable-interval schedule decreased signal detection's measure of discrimination but left its measure of response bias and the matching law measures unchanged. Data collected only until a subject's first changeover response following presentation of a long or a short tone showed higher values for both methods' measures of discrimination, no change in signal detection's measure of response bias, and lower values for the matching law's measure of response bias. Relationships between the matching law's and signal detection's methods of analyzing choice are discussed. It is concluded that a signal detection analysis is more efficient for examining changes in the difficulty of a discrimination, whereas a matching law analysis is more effective for examining the effects of changes in relative reinforcer frequency.