Edward L. Wike

Comments on Clark's “The language-as-fixed-effect fallacy”

Clarks arguments for treating language materials as random rather than fixed effects are examined, and the problems with random effects designs and approximate statistical tests (quasiF-ratios) are reviewed. In view of the difficulties with Clarks recommended procedures and the present lack of knowledge regarding approximate tests, it is suggested that researchers use fixed factors, which are better understood statistically, and seek nonstatistical generality by means of replication.

Effects of Alternating Partial Reinforcement and Alternating Delay of Reinforcement on a Runway Response

A replication and extension of the Cogan and Capaldi study was performed, using 60 rats. As in the original study, Ss with alternating reinforcement and nonreinforcement displayed patterned running during runway acquisition; Ss with alternating immediate and delayed reward did not. Additional groups with longer confinement periods exhibited patterned running under both alternating partial and alternating delay of reinforcement. The extinction results were in the opposite direction to those of Cogan and Capaldi; in all comparisons the partially delayed groups ran faster than the corresponding delayed groups.

Runway Performance as a Function of Delayed Reinforcement and Delay-Box Confinement

Two experiments were carried out to determine the effect of 20-sec. delay of reinforcement on 0, 33, 67, 100% of runway trials and 20-sec. confinement in the delay box on 0, 33, 67, and 100% of the extinction trials. Experiment I (N=76 rats) had 72 training and 30 extinction trials with 1 trial per day; Experiment II (N=47) had the same numbers of trials but 6 trials per day. A control group had 33% delay trials in a single block. The 100% delay group was significantly slower in training. The group with 67% delay in training ran the fastest in extinction. Running speed in extinction was inversely related to percentage of confinement in extinction. Shifts to higher percentages of confinement in extinction led to faster extinction; shifts to lower percentages of confinement led to slower extinction. The results, which were similar in the two experiments, were related to previous studies and discussed in terms of cognitive dissonance theory and a relativistic conception of reward.

The reward value of getting out of a starting box: Further extensions of Zimmerman’s work

SummaryFour control groups without secondary reinforcement training were tested in a Zimmerman Skinner box-runway apparatus. Two control groups, which were permitted to leave the box after bar pressing, pressed the bar as often as an experimental group with Sr training. The other two control groups, which were confined in the goal box after bar pressing, made significantly fewer bar presses. Zimmerman’s Sr interpretation of his results was questioned and the reward value of getting out of the starting box after bar pressing was discussed.

Sequences of reward magnitude and runway performance

Three groups of rats underwent 24 days of training and 12 days of extinction (three trials per day) in a runway under conditions of increasing (I), decreasing (D), and random (R) sequences of reward magnitudes (0, 45, and 500 mg). The I Ss ran faster over the daily trials, the D Ss slowed down, and the R Ss ran at approximately equal speeds on each trial. The patterned running observed in training persisted in extinction, with the R Ss running fastest and the I Ss next. The results were discussed in terms of Capaldi’s sequential theory and Amsel’s frustration theory.

DURATION OF DELAY, DELAY-BOX CONFINEMENT, AND RUNWAY PERFORMANCE

Rats (N=69) were trained in a runway under 0-, 10-, and 20-sec. delays and extinguished with 0-, 10-, 20-sec. delay-box confinements using one trial/day in each stage of the experiment. Duration of delay in training was inversely related to running speed in acquisition and in extinction. Duration of confinement in extinction was inversely related to running speed in extinction. Increases in delay from training to extinction led to faster extinction; decreases in delay to slower extinction.

Presentation rates and verbal discrimination learning

The effects of 1-, 2-, 4-, and 8-sec anticipation times and 1-, 2-, 4-, and 8-sec study times on verbal discrimination learning were investigated using 192 college students. In Experiment 1 (96 Ss), the items were more meaningful and there were fewer pairs of items than in Experiment 2. In Experiment 1, anticipation time alone was significant, with better performance resulting from a longer anticipation time. In Experiment 2 (96 Ss), both study time and anticipation time were significant; with longer times, better performance was found. The results were discussed in terms of frequency theory.

SELECTIVE LEARNING AS A FUNCTION OF DIFFERENTIAL CONSUMMATORY ACTIVITY

Recent research (4, 5 ) on the mechanism of reinforcement has resulted in a stress upon the amount of evoked consummatory activity as a primary determiner of the reinforcement value of an incentive. Spence (6) has emphasized the importance of the classically conditioned fractional antedating goal reaction as the hypothetical mechanism underlying incentive motivation, K, and that K in multiplicative combination with habit strength, H, produces excitatory tendency, E. It is readily apparent that amount of reward studies (1, 2, 7 ) involve a. number of confounded factors. For example, a large incentive results not only in a greater amount of food, but also in a longer consumption time, a possible different rate of consumption, possibly greater secondary reinforcement from the perceived size of the incentive, more separate consummatory responses when the food is in pellet form, etc. Likewise studies of duration of consummatory activity (3) are confounded by the amount of incentive and possibly other factors. The purpose of the present study was to answer the following question: Can selective learning occur as a consequence of differential dzcrations of consummatory activity when the amount of reward is held constant? Thirsty rats were given different drinking periods at the end boxes of a T-maze but allowed to consume the same amount of water. Two different sizes of drinking tubes were used. The experiment controlled the factor of amount of reward which covaried with the duration of consummatory response in investigations such as Festingers ( 3 ). It should be recognized, however, that in the present experiment the rate of water intake and vigor of the goal response varied with the drinking tube size.

Habit Reversal as a Function of Length of Partial Delay of Reinforcement

While the effects of delay of reinforcement on response acquisition are well known ( 6 ) , there is comparatively little information regarding its effects upon resistance to extinction. Recently, Fehrer ( 3 ) has demonstrated that a constant delay of reinforcement produces an instrumental response that is more resistant to extinction than when reward is immediately forthcoming. In the case of partially delayed reinforcement, in which delay of reward is imposed on a fraction of the training trials, there is considerable evidence (1, 8, 9, 10, 11) that delay of reinforcement enhances resistance to extinction. In addition Logan, et al. ( 7 ) have shown that that effect of partially delayed reinforcement is directly related to the length of delay. When the magnitude of partial delay was increased to 30 sec., the partially delayed Ss ran more rapidly during extinction than control Ss with either no delay or a partial delay of 9 sec. Since all these studies of partially delayed reinforcement have used runways, it appears desirable to investigate the generality of some of these findings in a more complex situation. In the present study a selective learning situation, a single unit T-maze, is employed, and the time required to achieve reversal in a position response constitutes the index of response strength. Some advantages of this criterion of response strength have been pointed out elsewhere ( 5 ). The purpose of the present investigation is then to determine whether or not the acquisition and reversal of a T-maze position response are affected by the length of partially delayed reinforcement.

The robustness of homogeneity of variance tests for asymmetric distributions: A Monte Carlo study

Robustness properties of three equality of variances tests, the jackknife procedure, the k-sample Box-Andersen test, and Levene’s z test, were investigated and compared in a Monte Carlo study employing small samples from underlying gamma distributions and gamma mixtures. Overall, the jackknife test yielded rejection proportions usually closer to nominal levels under departures from normality, but the Box-Andersen test performed relatively well when departures from normality were not extreme. Four recommendations regarding testing variances were offered.