Gary B. Nallan

Feature-positive and feature-negative learning in the rhesus monkey and pigeon.

In separate experiments four monkeys and eight pigeons were presented with displays containing one red and two green keys and displays containing three green keys. During feature-positive phases, responses to displays containing the one red and two green keys were reinforced on a fixed-ratio schedule, while responses to displays containing the three green keys were never reinforced. During feature-negative phases, only responses to the three green key displays were reinforced. For monkeys in Experiment 1, both between and within subject analyses indicated that the learning of feature-positive discriminations was superior to the learning of feature-negative discriminations. The within subject analysis further revealed that performance on a feature-positive discrimination was retarded following exposure to a feature-negative discrimination, while performance on a feature-negative discrimination was enhanced following exposure to a feature-positive discrimination. Experiment 2 replicated the essential aspects of these reversal effects in four experimental pigeons. Evidence that these reversal effects were not simply a function of time was provided by four control birds exposed to only a feature-positive or feature-negative discrimination.

Transfer effects in feature-positive and feature- negative learning by adult humans

In two experiments, college students performed a feature-positive or a feature-negative discrimination task based on colors or symbols and were then transferred to a feature-positive or feature-negative discrimination based on the other stimulus dimension (symbols-colors, colors-symbols). Initial task results yielded a substantial feature-positive effect and indicated that the color task was easier than the symbol task. Transfer task results indicated that the feature-positive effect was maintained and showed that consistent transfer (positive-positive, negative-negative) led to superior performance on the transfer task. These results were obtained when the correct solution to the initial task was provided to the subject prior to transfer (Experiment 1) and when it was not (Experiment 2). These results systematically replicated the existence of the feature-positive effect in adult humans and showed that both feature-positive and feature-negative discrimination learning were facilitated by consistent examples of these problems.

Some effects of rotation and centrifugally produced high gravity on taste aversion in rats

Rats were exposed to a single pairing of grape juice and rotation or rotation plus high gravity (5 g or 10 g). They were then tested separately for grape juice consumption over the next several days. High gravity did not reduce the amount of grape juice consumed, and nonrotated controls drank significantly more throughout testing. A high resistance to extinction was evidenced by the fact that drinking suppression was maintained for the experimental groups throughout five tests.

Categorical shape and color coding by pigeons.

Categorical coding is the tendency to respond similarly to discriminated stimuli. Past research indicates that pigeons can categorize colors according to at least three spectral regions. Two present experiments assessed the categorical coding of shapes and the existence of a higher order color category (all colors). Pigeons were trained on two independent tasks (matching-to-sample, and oddity-from-sample). One task involved red and a plus sign, the other a circle and green. On test trials one of the two comparison stimuli from one task was replaced by one of the stimuli from the other task. Differential performance based on which of the two stimuli from the other task was introduced suggested categorical coding rules. In Experiment 1 evidence for the categorical coding of sample shapes was found. Categorical color coding was also found; however, it was the comparison stimuli rather than the samples that were categorically coded. Experiment 2 replicated the categorical shape sample effect and ruled out the possibility that the particular colors used were responsible for the categorical coding of comparison stimuli. Overall, the results indicate that pigeons can develop categorical rules involving shapes and colors and that the color categories can be hierarchical.

Transfer effects in feature-positive and feature-negative learning by pigeons

Pigeons were exposed to a simultaneous feature-positive or feature-negative symbol task and were then transferred to a simultaneous featurepositive or feature-negative color task. Four transfer sequences resulted: positive-positive, positive-negative, negative-positive, negative-negative. The initial task data indicated that a typical feature-positive effect can be obtained in pigeons with a simultaneous task. Pigeons that were transferred to the feature-positive color task learned very rapidly, regardless of whether their initial task experience was feature-positive or feature-negative. A ceiling effect may have masked differential transfer. Pigeons that were transferred to the feature-negative color task were influenced by their initial task experience, i.e., the negative-negative group performed significantly better on the transfer task compared with the positive-negative group.

Temporal Parameters of the Feature Positive Effect

Trial duration and intertrial interval duration were parametrically varied between groups of pigeons exposed to a discrimination involving the presence vs. the absence of a dot. Half the groups received the dot as the positive stimulus (feature positive groups) and half the groups received the dot as the negative stimulus (feature negative groups). Faster learning by the feature positive birds (feature positive effect) was found when the trial duration was short (5 sec) regardless of whether the intertrial interval was short (5 sec) or long (30 sec). No evidence for a feature positive effect was found when the trial duration was long (30 sec) regardless of the length of the intertrial interval (30 sec or 180 sec). The results suggest that short trial duration is a necessary condition for the occurrence of the feature positive effect, and neither intertrial interval nor trial duration/intertrial interval ratio are important for its occurrence. The suggestion that mechanisms underlying the feature positive effect and autoshaping might be similar was not supported by the present experiment since the trial duration/intertrial interval ration parameter appears to play an important role in autoshaping but not the feature positive effect.

Generalization gradients following differential intradimensional autoshaping.

Three pigeons were trained on a differential, intradimensional autoshaped discrimination. A 45° line tilt was always paired with food whereas a 15° line tilt was never paired with food. All subjects learned the discrimination within 17 sessions. The pigeons were then given generalization tests in extinction over seven line tilts (0°, 15°, 30°, 45°, 60°, 75°, and 90°). The subjects yielded generalization gradients with maxima at 45° and minima at 15°. An area shift, but no peak shift, was found for each subject.

Identity Relation Can Serve as the Distinguishing Feature in Feature-Positive and Feature-Negative Learning Research

In three experiments, adult humans experienced a feature-positive or featurenegative letter discrimination task. The distinctive feature was a match in letters on one side of the card, i.e., the identity relation. In Experiment 1, the matching letters were next to each other, or separated by one other letter, or separated by two other letters. Feature-positive persons solved the task in significantly fewer trials than feature-negative persons. In Experiment 2, separate groups of feature-positive and feature-negative persons performed tasks where the matching letters were next to each other, or separated by one other letter, or separated by two other letters. Feature-positive persons solved the task in significantly fewer trials, and there was a trend toward faster learning when the matching letters were closer together. In Experiment 3, separate groups of feature-positive and feature-negative persons performed tasks where the matching letters were separated by one letter or were separated by three letters. Feature-positive persons solved the task significantly faster than feature-negative persons, and there was significantly faster learning when the matching letters were closer together. These results indicate that the identity relation can serve as a distinctive feature in visual discrimination problems.

Feature-positive effect in adults and attention to portion of stimulus array

In three experiments, adult humans were tested in a feature-positive or feature-negative simultaneous symbol task. In Experiment 1, some persons focused on the correct side of the stimulus cards, whereas other persons focused on the not-correct side of the stimulus cards. The feature-positive group learned faster than the feature-negative group did in the correct side condition; the feature-negative group learned faster than the feature-positive group did in the not-correct side condition. In Experiments 2 and 3, all persons focused on both the correct and not-correct sides of the stimulus cards. Under these circumstances, feature-positive and feature-negative performances were comparable. These results indicated that the usual superiority of feature-positive over feature-negative learning results from a tendency to attend to only a portion of the stimulus array.

The role of elicited responding in the feature-positive effect

Hearst and Jenkins proposed in 1974 that elicited responding accounts for the feature-positive effect. To test this position, pigeons were exposed to a feature-positive or feature-negative discrimination between successively presented displays--one consisted of a red and a green response key and the other consisted of two green response keys. There were four main conditions: 5-5 (5-sec trials, 5-sec intertrial intervals), 5-30, 30-30, and 30-180. Conditions 5-30 and 30-180 should produce the largest amount of elicited responding, and therefore the largest feature-positive effects. A response-independent bird was yoked to each response-dependent bird to allow direct assessment of the amount of elicited responding generated by each condition. Contrary to the predictions by Hearst and Jenkinss theory, response-dependent birds showed large feature-positive effects in each condition. The largest feature-positive effect was obtained in condition 5-5. Response-independent birds produced similar results, but manifested low response rates.