Eliot Shimoff

An Experimental Analysis of Rule-Governed Behavior

Contingency-shaped behavior is behavior directly controlled by the relations between responses and their consequences. But behavior may also come under the control of antecedent stimuli, stimuli in the presence of which responses produce their consequences. We find important examples of such stimuli in human verbal communities, which arrange contingencies that bring behavior under the control of antecedent verbal stimuli called commands, instructions, or rules.

Effects of Recording Attendance on Grades in Introductory Psychology.

We recorded attendance for 57 students in an introductory psychology class by having them sign in at each class meeting. For the remaining 57 students, we counted the number of students attending, but kept no record of individual attendance. Students who signed in attended classes more often (absenteeism decreased by one third), and their grades on weekly multiple-choice quizzes were higher, even on questions based on material covered in the text but not in lectures. Thus, simply recording attendance (without awarding course credit for attendance) increased both attendance and overall academic performance.

Post-session verbal reports and the experimental analysis of behavior.

Experimental analyses of the performance of verbal subjects often include verbal reports, obtained during post-session interviews, about within-session covert verbal behavior (e.g., hypotheses about the contingencies). But such post-session reports are not necessarily accurate, and procedural details of how the samples were obtained are typically inadequate. Even when the post-session reports are accurate, the within-session hypotheses do not have the status of causes of within-session nonverbal performance. In an experimental analysis, it is important to treat such reports as instances, not causes, of behavior.

The Verbal Governance of Behavior

The above quotation can be interpreted to mean that some research practices are verbally governed and others are not. The topic of verbal governance is mainly concerned with how human behavior depends on its verbal antecedents. The language in which the topic of verbal governance has usually been framed is fraught with difficulties and inconsistencies (see Catania, Chapter 13). For example, verbal governance has often been discussed under the rubric of rule-governed behavior (e.g., Skinner, 1969). Yet in many disciplines outside of behavior analysis, rules are not regarded as instances of verbal behavior, and issues of verbal governance are often couched in terms of awareness or of distinctions between implicit and explicit learning (e.g., Reber, 1976).

The experimental analysis of verbal behavior.

Theverbal behavior literature, inthepages ofthis journal andelsewhere, includes both data-based research onverbal behavior and talk about verbal behavior. Acknowledging that wehaveourselves occasionally contributedtotheimbalance, wenevertheless believe that there should bemoreoftheformer andless ofthelatter. Forexample, weonce submitted achapter toabookabout verbally governed behavior, andwhenthebookappeared weweresurprised todiscover that amongthe10contributions initourswas theonlyonetopresent original data (Catania, Shimoff, & Matthews, 1989). We therefore offer here afewsuggestions about somepossibilities forenhancing theexperimental database fortheanalysis ofverbal behavior. Depending ontheavailable time andresources, wemaysomedaygettoone oranother ofthese topics ourselves. Butwe arenotlikely tostart onanyoftheminthe nearfuture, andwesurely will never beable tocoverallofthem, soweoffer these suggestions inthehopethat someothers will take themup.Theworst that wouldhappen ifmorethanoneinvestigator gottoworkon thesametopic isthateachwouldbecome moreexperienced ingathering dataonverbalbehavior andthat somefindings would bestrengthened (orquestioned) byreplication. Themostwelcome commentary we could getonoursuggestions wouldbethe presentation ofdatasets fromoneormore ofthese research lines. Verbally governed andcontingency-shaped

Using Computers to Teach Behavior Analysis

When it is impractical to provide behavior analysis students with extensive laboratory experience using real organisms, computers can provide effective demonstrations, simulations, and experiments. Furthermore, such computer programs can establish contingency-shaped behavior even in lecture classes, which usually are limited to establishing rule-governed behavior. We describe the development of computerized shaping simulations and the development of software that teaches students to discriminate among reinforcement schedules on the basis of cumulative records.

Temptation and the Maintenance of Trust

Levels of trust, defined by how far ahead in earnings each dyad member risked putting the other, were continuously monitored during two-party exchange. Two experiments investigated whether a bilateral punishment capability would enable trust to be maintained when a strong incentive to be untrustworthy (temptation) was introduced. In Experiment I (total subtraction), both persons were provided with a “start over” button that could set the other persons session earnings to zero. In Experiment II (partial subtraction), the start over button was replaced by a “subtraction” button that subtracted ten points from the others earnings. All pairs had previously demonstrated either unwillingness to trust or untrustworthy behavior during temptation. Experiment I found that the availability of the total substraction option allowed trust to be maintained in four of six pairs. In Experiment II, partial subtraction had similar effects for three of six pairs. A bilateral punishment capacity may thus facilitate the maintenance of trust under conditions that otherwise produce distrust and exploitation.

Stimulus intrusion on fixed-interval responding in the rat: The effects of electric shock intensity, temporal location, and response contingency

Abstract After a scalloped lever-press response pattern had developed under a fixed-interval food reinforcement schedule, a 15-sec electric shock was intruded for different groups of rats in the first, second, third, or fourth quarter of each inter-reinforcement interval. Shock intensity was systematically increased for individual rats over 70 sessions, from 0.05 to 1.6 mA. Additional between-groups comparisons involved response-dependent versus clock-dependent fixed-interval schedules, and response-dependent versus response-independent electric shock intrusion. Response rates within each fixed interval prior to, during, and following electric shock intrusion showed regular and reproducible increases and decreases under systematic application of the experimental variables. These results provide further evidence that the functions of a stimulus are determined in part by the parameters of intensity, response contingency, and temporal location with respect to reinforcement.

Human Responding on a Temporally Defined Schedule of Point-Loss Avoidance

Two experiments extended the investigation of temporally defined avoidance schedules to human Ss responding on a schedule of point-loss avoidance. Experiment I found that superimposition of a td-correlated added stimulus had no effect on response rate or distribution, while response-cost contingency of 1 point per response produced a decrease in response rate and an increase in failures to avoid 10-point losses. Discriminated responding was produced only by combination of the added stimulus and the response-cost contingency. Experiment II found that a high-effort response mani-pulandum was functionally equivalent to the response-cost contingency in behavioral effects. The results are consistent with Weiner’s explanation for apparent insensitivity of human responding to nonresponse contingency events and support the generic use of the term “cost” in exchange theories.

Measurement of Behavioral Effects of the CER Procedure

While the disadvantages of absolute measures of the behavioral effects of response rate in the CER procedure are well-known, there are several often unrecognized difficulties with the use of suppression ratios, the most serious relating to changes in pre-S1 response rates and to non-uniformity of response rate in the presence of S1. Several alternative metrics either describing the response pattern in S1 or focusing on some specific component of the pattern are suggested.