Optimization and the matching law as accounts of instrumental behavior

The interaction between instrumental behavior and environment can be conveniently described at a molar level as a feedback system. Two different possible theories, the matching law and optimization, differ primarily in the reference criterion they suggest for the system. Both offer accounts of most of the known phenomena of performance on concurrent and single variable-interval and variable-ratio schedules. The matching law appears stronger in describing concurrent performances, whereas optimization appears stronger in describing performance on single schedules.     

Choice, experience, and the generalized matching law

Five pigeons were exposed to different pairs of concurrent variable-interval, variable-interval schedules on nine experimental conditions of 30 sessions each. For every session, the parameters of the generalized matching equation were computed for the first five, six, seven, eight, and nine experimental conditions. The exponent a, both for response and time distribution, tended to decrease with increases in number of experimental conditions and to increase with number of sessions per condition, but values of k (bias) varied unsystematically. When the subjects were exposed to five new pairs of schedules, with 55 sessions per condition, the findings were confirmed. Data from the literature on the generalized matching law suggest that the variability of exponent values may be explained in part by the use of naive or experienced subjects in different investigations and by the variability in number of experimental conditions and in number of sessions per condition.     

Local contrast in behavior allocation during multiple-schedule components

Allocation of responses between two keys was studied during two alternating multiple-schedule components. Responses were recorded in successive quarters of each component. Variable-interval reinforcer schedules on the two keys were constant throughout the experiment for one (constant) component and were varied over conditions on one key for the other, producing changes in reinforcer ratios for the varied component. Behavior allocation for the first quarter of the constant component was inversely related to varied-component reinforcer ratios, a form of local contrast, but this relationship was not observed later in the component. During the first quarter of the varied component, slopes of matching lines were high and decreased later in the component. It is argued that this form of local contrast cannot be explained in terms of reallocation of extraneous reinforcers between components, and that the matching law for concurrent operants does not capture some sources of control over behavior allocation. A simple extension of the matching law is offered that adequately describes behavior changes during both components. A version of this formulation can predict contrast effects in absolute response rates.     

The effects of concurrent responding and reinforcement on behavioral output

Four birds key pecked on concurrent variable-interval one-minute variable-interval four-minute schedules with a two-second changeover delay. Response rates to the variable-interval one-minute key were then reduced by signaling its reinforcer availability and later by providing its reinforcers independently of responding. Each manipulation increased response rates to the variable-interval four-minute key even though relative reinforcement rates were unchanged. In a final phase, eliminating the variable-interval one-minute key and its schedule produced the highest rates of all to the variable-interval four-minute key. These results show that both reinforcement and response rates to one schedule influence response rates to another schedule. These results join those of Guilkey, Shull, & Brownstein (1975) in failing to replicate Catania (1963). Moreover, they violate the predictions of the equation for simple action (de Villiers & Herrnstein, 1976). In terms of a median-rate measure (reciprocal of the median interresponse time), rates to the variable-interval four-minute key were high when responding was not reduced to the variable-interval one-minute key and were low when it was reduced. This rate difference suggests a process difference between concurrent-schedule procedures that maintain high concurrent response rates versus those that do not. This process difference jeopardizes attempts to integrate single- and concurrent-operant performances within a single formulation.     

Changeover behavior and preference in concurrent schedules

Pigeons were trained on a multiple schedule of reinforcement in which separate concurrent schedules occurred in each of two components. Key pecking was reinforced with milo. During one component, a variable-interval 40-s schedule was concurrent with a variable-interval 20-s schedule; during the other component, a variable-interval 40-s schedule was concurrent with a variable-interval 80-s schedule. During probe tests, the stimuli correlated with the two variable-interval 40-s schedules were presented simultaneously to assess preference, measured by the relative response rates to the two stimuli. In Experiment 1, the concurrently available variable-interval 20-s schedule operated normally; that is, reinforcer availability was not signaled. Following this baseline training, relative response rate during the probes favored the variable-interval 40-s alternative that had been paired with the lower valued schedule (i.e., with the variable-interval 80-s schedule). In Experiment 2, a signal for reinforcer availability was added to the high-value alternative (i.e., to the variable-interval 20-s schedule), thus reducing the rate of key pecking maintained by that schedule but leaving the reinforcement rate unchanged. Following that baseline training, relative response rates during probes favored the variable-interval 40-s alternative that had been paired with the higher valued schedule. The reversal in the pattern of preference implies that the pattern of changeover behavior established during training, and not reinforcement rate, determined the preference patterns obtained on the probe tests.     

Maximizing versus matching on concurrent variable-interval schedules

Maximization and matching predictions were examined for a time-based analogue of the concurrent variable-interval variable-ratio schedule. One alternative was a variable interval whose time base operated relatively independent of the schedule chosen, and the other was a discontinuous variable interval for which timing progressed only when selected. Pigeons switched between schedules by pecking a changeover key. The maximization hypothesis predicts that subjects will show a bias toward the discontinuous variable interval and undermatching; however the obtained results conformed closely to the predictions of the matching law. Finally, a quantitative comparison was made of the bias and sensitivity estimates obtained in published concurrent variable-interval variable-ratio analogue studies. Results indicated that only the ratio-based analogue of the concurrent variable interval variable ratio studied by Green, Rachlin, and Hanson (1983) produced significant bias toward the variable-ratio alternative and undermatching, as predicted by reinforcement maximization.     

Choice and segmented interreinforcement intervals

Pigeons were trained on a two-key concurrent schedule, where food reinforcers on one key were arranged by a simple variable-interval schedule and on the other key by a chain variable-interval variable-interval schedule. When the initial link of the chain was in effect, the pigeons tended to respond more on the simple variable-interval schedule, and hence less on the chain, than would be expected from a comparison of both the local and overall rates of reinforcement of the two schedules. When the terminal link of the chain was in effect, the pigeons responded more on the chain than would be expected from a comparison of the rates of reinforcement of the schedules then in effect. Overall responding on the chain was not proportional to overall reinforcement on the chain but rather was a by-product of responding during initial- and terminal-link phases.     

Drug discrimination under a concurrent fixed-interval fixed-interval schedule.

Pigeons were trained to discriminate 5.0 mg/kg pentobarbital from saline under a concurrent fixed-interval (FI) FI schedule of food presentation on which, after pentobarbital administration, responses on one key were reinforced with food under an FI 60-s component and responses on the other key were reinforced under an FI 240-s component. After saline administration, the schedule contingencies on the two keys were reversed. After both pentobarbital and saline, pigeons responded more frequently on the key on which responses had been programmed to produce the reinforcer under the FI 60 component of the concurrent schedule. The schedule was changed to concurrent FI 150 FI 150 s for drug-substitution tests. In each bird, increasing doses of pentobarbital, ethanol, and chlordiazepoxide produced increases in the proportion of responses on the key on which responses had been reinforced under the FI 60 component after pentobarbital administration during training sessions. The proportion of responses on that key was slightly lower for ethanol than for chlordiazepoxide and pentobarbital. At a dose of pentobarbital higher than the training dose, responding decreased on the key that had been reinforced under the FI 60 component during training sessions. Phencyclidine produced less responding on the key programmed under the FI 60-s component than did pentobarbital. Methamphetamine produced responding primarily on the key on which responses had been reinforced under the FI 60-s component after saline administration.     

DEVIATIONS FROM MATCHING AS A MEASURE OF PREFERENCE FOR ALTERNATIVES IN PIGEONS

Preferences for larger or smaller formally defined response classes were investigated in a concurrent schedule procedure. Twelve pigeons were run on a series of concurrent variable-interval reinforcement schedules, from which baseline matching functions were obtained. An experimental phase followed, in which a second response key was available in one concurrent schedule alternative. For half the birds, the second key was programmed identically with the first; for the other half, the added key was programmed for extinction, with position irrelevant. Comparison of baseline and experimental matching functions revealed no systematic changes in either slope or intercept for birds in the latter group. Systematic shifts in function intercepts in the former group indicate a response bias toward the response-constrained (single-key) schedule alternative. Although contrary to the literature of preference for choice, this finding may be interpretable through an account dealing with imposed variability of responding.     

Choice between response units: The rate constancy model

In a conjoint schedule, reinforcement is available simultaneously on two or more schedules for the same response. The present experiments provided food for key pecking on both a random-interval and a differential-reinforcement-of-low-rate (DRL) schedule. Experiment 1 involved ordinary DRL schedules; Experiment 2 added an external stimulus to indicate when the required interresponse time had elapsed. In both experiments, the potential reinforcer frequency from each component was varied by means of a second-order fixed-ratio schedule, and the DRL time parameter was changed as well. Response rates were described by a model stating that time allocation to each component matches the relative frequency of reinforcement for that component. When spending time in a given component, the subject is assumed to respond at the rate characteristic of baseline performance. This model appeared preferable to the absolute-rate version of the matching law. The model was shown to be applicable to multiple-response concurrent schedules as well as to conjoint schedules, and it described some of the necessary conditions for response matching, undermatching, and bias. In addition, the pigeons did not optimize reinforcer frequency.     

Choice performance in several concurrent key-peck treadle-press reinforcement schedules

Five pigeons were exposed to several concurrent variable-interval food reinforcement schedules. For three subjects, one component of the schedule required a key-pecking response, the other a treadle-pressing response. For the other two subjects, both schedule components required treadle-pressing responses. The relative probability of reinforcement associated with the manipulanda was varied from 0 to 1.0 in 13 experimental conditions for the Key-Treadle subjects and nine conditions for the Treadle-Treadle subjects. The results indicated that the logarithms of relative time spent responding, and the logarithms of relative number of responses emitted on a manipulandum, approximated direct linear functions of logarithms of the relative frequencies of reinforcement associated with that manipulandum. No systematic bias in favor of time spent key pecking over time spent treadle pressing was apparent for the Key-Treadle subjects. All subjects exhibited undermatching, in that the ratios of time and response allocation at the alternatives systematically differed from the ratios of reinforcers obtained from the alternatives in the direction of indifference. Key pecking appeared to have no special link to food beyond treadle pressing or what would be expected on the basis of the reinforcement dependencies alone.     

Matching and maximizing with concurrent ratio-interval schedules.

Animals exposed to standard concurrent variable-ratio variable-interval schedules could maximize overall reinforcement rate if, in responding, they showed a strong response bias toward the variable-ratio schedule. Tests with the standard schedules have failed to find such a bias and have been widely cited as evidence against maximization as an explanation of animal choice behavior. However, those experiments were confounded in that the value of leisure (behavior other than the instrumental response) partially offsets the value of reinforcement. The present experiment provides another such test using a concurrent procedure in which the confounding effects of leisure were mostly eliminated while the critical aspects of the concurrent variable-ratio variable-interval contingency were maintained: Responding in one component advanced only its ratio schedule while responding in the other component advanced both ratio schedules. The bias toward the latter component predicted by maximization theory was found.     

Sensitivity to relative reinforcer rate in concurrent schedules: independence from relative and absolute reinforcer duration

Twelve pigeons responded on two keys under concurrent variable-interval (VI) schedules. Over several series of conditions, relative and absolute magnitudes of reinforcement were varied. Within each series, relative rate of reinforcement was varied and sensitivity of behavior ratios to reinforcer-rate ratios was assessed. When responding at both alternatives was maintained by equal-sized small reinforcers, sensitivity to variation in reinforcer-rate ratios was the same as when large reinforcers were used. This result was observed when the overall rate of reinforcement was constant over conditions, and also in another series of concurrent schedules in which one schedule was kept constant at VI ached 120 s. Similarly, reinforcer magnitude did not affect the rate at which response allocation approached asymptote within a condition. When reinforcer magnitudes differred between the two responses and reinforcer-rate ratios were varied, sensitivity of behavior allocation was unaffected although response bias favored the schedule that arranged the larger reinforcers. Analysis of absolute response rates ratio sensitivity to reinforcement occurrred on the two keys showed that this invariance of response despite changes in reinforcement interaction that were observed in absolute response rates on the constant VI 120-s schedule. Response rate on the constant VI 120-s schedule was inversely related to reinforcer rate on the varied key and the strength of this relation depended on the relative magnitude of reinforcers arranged on varied key. Independence of sensitivity to reinforcer-rate ratios from relative and absolute reinforcer magnitude is consistent with the relativity and independence assumtions of the matching law.     

Concurrent schedules: Spatial separation of response alternatives

Four pigeons were exposed to independent concurrent variable-interval 20-second variable-interval 60-second schedules of reinforcement. A transparent partition was inserted midway between the two response keys. The length of the partition was systematically manipulated. Increasing partition length produced a decrease in changeover rate in Experiment 1. Over-matching was observed with a partition length of 20 centimeters. In Experiment 2 a four-second limited hold was added to the schedules. Increasing partition length produced a decrease in changeover rate that exceeded the decrease observed in Experiment 1. This manipulation produced nearly exclusive choice of the variable-interval 20-second component. The present results, together with results obtained in related research, suggest that deviation from matching is a function of procedural variables that determine the consequences of a changeover response.     

Preference for starting and finishing behavior patterns

Pigeon's key pecking was reinforced with food in two experiments in which the correspondence between preference for starting one of two reinforced behavior patterns and the likelihood of finishing it subsequently was examined. Reinforcers were scheduled according to concurrent schedules for two classes of interresponse times, modified such that reinforcers followed a center-key peck terminating either a shorter interresponse time started by a left-key peck or a longer interresponse time started by a right-key peck. In Experiment 1, the times when reinforcers potentially were available were not discriminated, whereas in Experiment 2 they were. Absolute reinforced pattern durations were varied. The relative frequency of starting a particular pattern was highly correlated with relative frequency of that completed pattern in both experiments. Other relations between starting and finishing a pattern depended on whether reinforced interresponse times were discriminated. For instance, preference for starting a pattern sometimes correlated negatively with the likelihood of subsequently completing it. The present experiments are described as capturing part of the ordinary language meaning of "intention," according to which an organism's behavior at one moment sets the occasion for an observer to say that the organism "intends" in the future to engage in one behavior rather than another.     

Melloration and maximization of reinforcement minus costs of behavior

Eight pigeons were exposed to independent concurrent schedules. Concurrent variable-interval 60-second variable-interval 60-second schedules were presented to one group of four subjects. Following baseline training, a limited hold was added to one of the schedules and the duration of the hold was decreased in successive conditions. Concurrent variable-interval 120-second variable-interval 40-second schedules were presented to another group of four subjects. These subjects were first exposed to decreasing durations of a limited hold in the variable-interval 40-second component. After replication of the baseline, a limited hold in the variable-interval 120-second component was decreased in duration. The initial durations of the holds were determined from the subjects' responding in the baseline conditions. A duration was chosen such that approximately 25% of the scheduled reinforcers would be canceled if responding remained unchanged.

Approximate matching of time proportions and reinforcement proportions was observed when the limited hold was added to the variable-interval 60-second schedule and when the limited hold was added to the variable-interval 40-second schedule. Time proportions were less extreme than reinforcement proportions when the limited hold operated in a variable-interval 120-second schedule. Overall reinforcement rates tended to decrease with continued training in concurrent schedules with a limited hold. Absolute deviations from time matching also decreased. The results provide evidence against the principle of reinforcement maximization, and support Herrnstein and Vaughan's (1980) melioration hypothesis.

     

Concurrent random-interval schedules and the matching law

In Experiment I, a group of eight pigeons performed on concurrent random-interval schedules constructed by holding probability equal and varying cycle time to produce ratios of reinforcer densities of 1:1, 3:1, and 5:1 for key pecking. Schedules for a second group of seven were constructed with equal cycle times and unequal probabilities. Both groups deviated from simple matching, but the two forms of the schedules appeared to produce no consistent patterns of deviation. The data were found to be consistent with those obtained in concurrent variable-interval situations. The parameters of the matching equation in the form of Y=k X^a were estimated; the value of k was unity and a was 0.84. In Experiment II, six pigeons were exposed to two conc RI RI schedules in which one component increasingly approximated an FI schedule. The value of k was not 1.0. Concurrent RI RI schedules were shown to represent a continuum from conc FI VI to conc VI VI schedules. The use of the exponential equation in testing “matching laws” suggests that a<1 will continue to be observed, and this will set limits on the form of new laws and the assumed or rational values of the component variables in these laws.     

Residence time and choice in concurrent foraging schedules

Five pigeons were trained on a concurrent-schedule analogue of the “some patches are empty” procedure. Two concurrently available alternatives were arranged on a single response key and were signaled by red and green keylights. A subject could travel between these alternatives by responding on a second yellow “switching” key. Following a changeover to a patch, there was a probability (p) that a single reinforcer would be available on that alternative for a response after a time determined by the value of λ, a probability of reinforcement per second. The overall scheduling of reinforcers on the two alternatives was arranged nonindependently, and the available alternative was switched after each reinforcer. In Part 1 of the experiment, the probabilities of reinforcement, ρ_red and ρ_green, were equal on the two alternatives, and the arranged arrival rates of reinforcers, λ_red and λ_green, were varied across conditions. In Part 2, the reinforcer arrival times were arranged to be equal, and the reinforcer probabilities were varied across conditions. In Part 3, both parameters were varied. The results replicated those seen in studies that have investigated time allocation in a single patch: Both response and time allocation to an alternative increased with decreasing values of λ and with increasing values of ρ, and residence times were consistently greater than those that would maximize obtained reinforcer rates. Furthermore, both response- and time-allocation ratios undermatched mean reinforcer-arrival time and reinforcer-frequency ratios.     

Nonstable concurrent choice in pigeons

Six pigeons were trained on concurrent variable-interval schedules in which the arranged reinforcer ratios changed from session to session according to a 31-step pseudorandom binary sequence. This procedure allows a quantitative analysis of the degree to which performance in an experimental session is affected by conditions in previous sessions. Two experiments were carried out. In each, the size of the reinforcer ratios arranged between the two concurrent schedules was varied between 31-step conditions. In Experiment 1, the concurrent schedules were arranged independently, and in Experiment 2 they were arranged nonindependently. An extended form of the generalized matching law described the relative contribution of past and present events to present-session behavior. Total performance in sessions was mostly determined by the reinforcer ratio in that session and partially by reinforcers that had been obtained in previous sessions. However, the initial exposure to the random sequence produced a lower sensitivity to current-session reinforcers but no difference in overall sensitivity to reinforcement. There was no evidence that the size of the reinforcer ratios available on the concurrent schedules affected either overall sensitivity to reinforcement or the sensitivity to reinforcement in the current session. There was also no evidence of any different performance between independent and nonindependent scheduling. Because of these invariances, this experiment validates the use of the pseudorandom sequence for the fast determination of sensitivity to reinforcement.