Stagewise multidimensional visual discrimination by pigeons

We trained six pigeons in a stagewise Multiple Necessary Cues (MNC) go/no‐go task to document the dynamics of discrimination learning involving increasingly complex visual stimuli. The compound stimuli were composed from four dimensions, each of which could assume either of two extreme values or their intermediate value: Shape, Size, Line Orientation, and Brightness. Starting with a stimulus composed entirely from intermediate values, we replaced those values with each of the two extreme dimensional values in four successive stages, thereby increasing the stimulus set from 2 in Stage 1 to 16 in Stage 4. In each stage, only one combination of values signaled food (S⁺), whereas the remaining combinations did not (S^?s). We calculated the rate of pecking during the first 15 s of each stimulus presentation and, in any given stage, training continued until the rate of responding to all of the S^?s was less than 20% of the rate of responding to the S⁺. All pigeons successfully acquired the final discrimination, suggesting that they attended to all of the dimensions relevant for the discrimination. We also replicated the key results of prior MNC studies: (1) the number of extreme dimensional values in each stage was positively related to the amount of training required for pigeons to acquire the discrimination; (2) attentional tradeoffs were most often observed when three or four dimensions were being trained; and (3) throughout training, the number of dimensional values in the S^?s that differed from the S⁺ was positively related to their discriminability from S⁺.     

Asymmetrical stimulus control and the mirror-image problem

An analysis of the mirror-image discrimination literature revealed a number of inconsistent and paradoxical results. A unifying conception was proposed and it was argued that lateral mirror-image stimuli constitute a special class of problems on which effective performance is related to the introduction of an asymmetrical event into the task. Some implications were derived and tested in two experiments. First, five different response modes were compared for children required to discriminate two lateral or up-down mirror-image stimulus pairs. Acquisition was significantly faster on the latter problems. On the former, it was shown that responding to just one side of the lateral mirror-image stimuli led to significantly faster learning. Subsequent generalization tests indicated control by one part of the stimulus display for those who responded to one side. Next, Ss unable to learn the problem in Experiment I were given further acquisition trials. Only those children given training designed to promote selective stimulus control learned the problem.     

Generalization gradients following two-response discrimination training

Stimulus generalization was investigated using institutionalized human retardates as subjects. A baseline was established in which two values along the stimulus dimension of auditory frequency differentially controlled responding on two bars. The insertion of the test probes disrupted the control established to the two S^Ds during training. The discrimination was recovered between each test probe and the resulting gradients were stable across 10 test sessions. These gradients, supported by other two-response generalization studies, indicate that this type of two-response discrimination training divides the stimulus dimension into two functional classes separated by a region of transition from one class to the other. Each stimulus value in a class, which extends from an S^D outward to the functional limit of the dimension, controls a similar proportion of the two responses as each other value in the class. All values on the stimulus dimension control identical response rates with an absence of the usual generalization decrement. The latency of the initial response, however, shows a bimodal gradient with the modes at the S^D values.     

A developmental study of differences in initial coding and recoding of hypothesis information

Coding and recoding of hypothesis information was studied in children in Grades 3, 5, and 7, and college students. Each S was tested on several two-choice discrimination problems. A problem consisted of two trials, a training trial and a test trial. The information available for coding on the first trial was varied in three series of problems by using pairs of stimuli differing in 3, 4, or 6 dimensions. Recoding after an error was facilitated in half of the problems in each series by allowing the stimuli to persist briefly after response and feedback. The stimuli terminated immediately after response and feedback in the remaining problems. The second trial of a problem, in which the stimuli differed on only one dimension, was used to probe for the information coded or recoded on the first trial. On the average, fewer dimensions were coded per trial by the younger Ss than by the older Ss. Neither the effects of negative feedback nor the effects of stimulus persistence were found to be related to developmental level. It was concluded that a deficiency in young children's ability to code, but not recode, hypothesis information contributes to their less efficient performance in discrimination problems.     

Effects of stimulus differentiation training upon hypotheses,strategies, and stereotypes in discrimination learning among kindergarten children

Two groups of kindergarten children received stimulus differentiation training either with feedback (experimental Ss) or without (controls), prior to presentation of a series of discrimination-learning problems using blanktrial probes. Results were as follows: Performance of both groups was virtually perfect during differentiation training (less than one error per 1000 trials). In transfer, however, experimental Ss solved more problems, generated more consistent Ss during blank-trial probes, manifested fewer response sets, were more likely to abandon a disconfirmed H, sampled more locally consistent Ss following errors, and retained a confirmed H more frequently than control Ss. Experimental Ss also manifested strategy Sys (hypothesis sampling systems) predominately, while controls generated only stereotypes. Findings were discussed in relation to theoretical perspectives derived from Piagetian theory and developmental learning theory.     

On the discriminative control of concurrent responses: the relations among response frequency, latency, and topography in auditory generalization

Human subjects were used in a study of auditory generalization following multiple-response discrimination training. The relations observed among stimulus intensity, response probability, and response latency were invariant with respect to whether the two vocal responses conditioned were topographically discrete, as in one experiment, or topographically continuous, as in another. The major findings were:

1. The probabilities associated with a specific response were maximal over several stimulus values at the extreme ends of the continuum, then dropped sharply at stimuli intermediate to the initial S^D's as the probability of the alternative response increased.

2. Overall response latency was inversely related to the relative frequency of the two responses at each stimulus value. When the two responses were most nearly equal in probability, latencies were maximal; when one response had close to unit or zero probability, latencies were minimal.

3. Analysis of the latencies of the two responses, taken separately, revealed: (a) an increase in latency as the difference between the test stimulus and the initial S^D increased; (b) a sharp discontinuity in the latency gradient and reversal in trend at intermediate stimulus intensities; and (c) at a given stimulus value, latencies associated with the stochastically dominant response were consistently shorter than those of the nondominant response.

4. No changes in response topography (fundamental frequency) were correlated with the characteristic changes in probability and latency during stimulus generalization.

     

“Generalized imitation” on a nonimitative task: Effects of modeling and task history

Two experiments investigated characteristic generalized imitation procedures on a nonimitative successive visual discrimination task. In Experiment I, no discriminative behavior was obtained though a number of procedures designed to enhance discrimination were employed. The introduction of a differential modeling procedure after nine or 10 sessions of nondifferential performance was ineffective in producing differential responding for three of four Ss. In the second experiment, the differential modeling procedure was introduced as an initial manipulation and was successful in establishing discriminative performance in all Ss. After several sessions of differential responding, the observation of a model performing nondifferentially was relatively ineffective in altering the pattern of responding. These results suggest (1) that modeling may be functional in altering a S's behavior depending on what point in the procedures the manipulation is introduced and (2) that certain procedural variables may define a functional response class characterized by the failure to develop discriminative responding.     

An analysis of the solution of PAN ambiguous-cue problems by rhesus monkeys

Twenty-four monkeys were given 2-choice discrimination problems composed of three planometric plaque stimuli: P, the rewarded stimulus; N, the nonrewarded stimulus; and A, the ambiguous stimulus which was negative when paired with P, but positive when paired with N. When both pair of stimuli, PA and NA, were presented within a given session S was forced either to approach or to avoid the A plaque depending upon the stimulus with which it was paired. The results corroborated previous reports for plaque stimuli by showing PA performance to be superior to NA. The data also revealed that when plaque stimuli with distinctive cues were employed, NA performance exceeded PA as previously reported for stereometric object stimuli. Pretraining Ss with plaque stimuli possessing distinctive cues, then switching to plaques with less distinctive cues, also resulted in superior NA performance. These findings are discussed in the context of an interference-cue theory.     

Effect of prior Pavlovian discrimination training upon learning an operant discrimination

The effect of Pavlovian discrimination training with two stimuli upon subsequent learning of an operant discrimination involving those stimuli was studied. After preliminary lever press training, the lever was removed and thirsty rats received noncontingent pairings between S₁ (a tone or a clicker) and water reinforcements, whereas S₂ (a clicker or a tone) occurred always without reinforcement. This procedure presumably established S₁ as a positive CS for respondent behavior, whereas S₂ was established as an inhibitory CS. Following this training, the lever was reintroduced and the rats were trained on an operant (lever pressing) discrimination involving S₁ and S₂. For the Consistent Ss, S₁ was the S^D and S₂ the S^Δ in the operant discrimination; for the Reversed Ss, S₂ served as S^D and S₁ as S^Δ. The Consistent Ss learned the operant discrimination significantly faster than did the Reversed Ss. The result emphasizes the importance of respondents, conditioned to S^D and S^Δ, which modulate operant performance to these stimuli.     

Extradimensional shift learning in pigeons: Positive and negative transfer produced by prior discrimination training

Four groups of pigeons were trained on a free operant successive discrimination between line-stimuli differing in orientation. The groups differed according to their previous treatment. Two groups had received true discrimination (TD) training with stimuli differing in colour and two groups had received pseudo-discrimination (PD) training. For one pair of groups the line-stimulus that was to become the positive stimulus in the transfer stage of the experiment was superimposed on both colours and in these the subjects given PD training learned the orientation discrimination more readily than those given TD training. In the other pair of groups TD animals learned more readily than PD. These results require us to modify current theories of attentional factors in transfer.     

Effects of stimulus similarity in discrimination training upon wavelength generalization in pigeons.

Thirty pigeons were given variable interval training to peck a 555-nm. light and then were tested for wavelength generalization. The subjects were later assigned to 1 of 3 groups, matched for both relative generalization slope and response rate. One group then received successive discrimination training between the 555-nm. stimulus (S+) and a vertical white line on a 555-nm. background (S minus); another group experienced the same S+ but a vertical white line on a black background as S minus. A third group received a comparable amount of single stimulus training with the 555-nm. value. On a second wavelength generalization test, the first group yielded greater sharpening of generalization than the second group, whereas the third group showed no change from Test 1. These results indicate that the sharpening of generalization gradients by discrimination training is directly related to the similarity of the discrimination training stimuli.     

Comparatives,logical structures,and encoding in a transitive inference task

Four- to five-year-old children were asked questions on length involving single or double comparatives in a transitive inference task. The number of comparatives varied within or across pairs. The pairs of sticks were colorcoded. Initial pair-wise discrimination training on four adjacent pairs from a five-stick array with only verbal feedback was followed by tests, without feedback, on all possible pairs. When both comparatives were used, Ss learned adjacent pairs faster and more often reached criterion than when only one comparative term was used in training. In testing, Ss were successful on transitive inference tests only when the double-comparative relation was used within pairs during training. Retraining with visual feedback increased the number of Ss solving the inference test only in this condition. The findings are discussed in terms of how children may use processes other than logical operations to make transitive inferences.     

Contrast and stimulus generalization following prolonged discrimination training

Different groups of pigeons received discrimination training in which the reinforcement-associated and extinction-associated stimuli were respectively either (a) a line tilt vs a blank key, (b) a blank key vs a line tilt, or (c) two different line tilts. The high response rates that developed to the positive stimulus in all groups during discrimination learning were maintained over 64 sessions of training. After these sessions, all subjects were tested for stimulus generalization along the line-tilt dimension. Gradients of relative (per cent) generalization around the stimulus associated with reinforcement (so-called excitatory gradients) and around the stimulus associated with extinction (so-called inhibitory gradients) were as steep as they typically are after much briefer training periods. These results do not support several of Terrace's predictions on the basis of the hypothesis that emotional responses develop to the stimulus associated with extinction during discrimination learning with errors, but eventually dissipate after extended training.     

Novelty as a reinforcer for position learning in children

The stimulus-familiarization-effect (SFE) paradigm, a reaction-time (RT) task based on a response to novelty procedure, was modified to assess response for novelty, i.e., a response-reinforcement sequence. Ninetysix kindergarten children were familiarized on one of two colored lights. Then Ss were confronted with a simple position discrimination task in which they could choose between two response buttons which produced either the previously familiarized light or a new (i.e., novel) light. Results showed that the choice response for the novel stimulus (NS) was facilitated by three variables: number of exposures to and duration per exposure of the familiar stimulus (FS), and degree of change provided by the NS relative to the FS. The parallel between the effective variables of this study and those which produce the SFE was noted, implicating an attention interpretation of the function of novelty as a reinforcer. The potential implications of attention for reinforcement theory and learning in general are discussed.     

Generalization and discrimination as a function of the SD — SΔ intensity difference

Three groups of four rats were trained on an auditory-intensity discrimination for 21 days. The S^D-S^Δ intensity difference for Group I was 10 db; for Group II, 20 db; and for Group III, 30 db. Following the initial discrimination training, the animals were tested for generalization of the bar-press response to seven novel S^Δ's which were presented intermingled with the original S^D and S^Δ values. Conclusions: (1.) The amount of simple discrimination training required to obtain fairly stable differences in S^D and S^Δ responding is an inverse function of the magnitude of the stimulus difference between S^D and S^Δ. (2.) Generalization gradients obtained immediately following simple discrimination training exhibit a maximum displaced from S^D in a direction also away from S^Δ. (3.) Gradients obtained following continued exposure to the multivalued S^Δ situation show a fairly stable maximum at the S^D value. (4.) Although the gradients tend to fall off systematically on either side of the continuum as distance from S^D is increased, they decrease most rapidly on the S^Δ limb of the gradient.     

Differential reinforcement of low rates performance by impulsive and reflective children

Third-grade boys classified as either cognitively impulsive or reflective were reinforced for key pressing according to a DRL (differential reinforcement of low rates) 6-sec schedule of reinforcement. Half of each group received instructions about the behavioral requirements for obtaining reinforcements. Prior to DRL training, impulsive Ss showed a low probability of key press responding at long interresponse time (IRT) intervals while reflective Ss exhibited an equal probability of terminating either short or long IRTs. During training and in the absence of instructions, impulsives exhibited a less precise temporal discrimination, characterized by a greater predominance of response bursts (0–2 sec IRTs) following reinforcements, than reflective Ss. While impulsive and reflective Ss displayed similar frequencies of collateral behavior between successively reinforced responses, impulsives engaged in the reinforced response more frequently and tended (p < .08) to obtain fewer reinforcements. Instructions served to enhance the DRL performance.     

Stimulus generalization, discrimination learning, and peak shift in horses.

Using horses, we investigated three aspects of the stimulus control of lever-pressing behavior: stimulus generalization, discrimination learning, and peak shift. Nine solid black circles, ranging in size from 0.5 in. to 4.5 in. (1.3 cm to 11.4 cm) served as stimuli. Each horse was shaped, using successive approximations, to press a rat lever with its lip in the presence of a positive stimulus, the 2.5-in. (6.4-cm) circle. Shaping proceeded quickly and was comparable to that of other laboratory organisms. After responding was maintained on a variable-interval 30-s schedule, stimulus generalization gradients were collected from 2 horses prior to discrimination training. During discrimination training, grain followed lever presses in the presence of a positive stimulus (a 2.5-in circle) and never followed lever presses in the presence of a negative stimulus (a 1.5-in. [3.8-cm] circle). Three horses met a criterion of zero responses to the negative stimulus in fewer than 15 sessions. Horses given stimulus generalization testing prior to discrimination training produced symmetrical gradients; horses given discrimination training prior to generalization testing produced asymmetrical gradients. The peak of these gradients shifted away from the negative stimulus. These results are consistent with discrimination, stimulus generalization, and peak-shift phenomena observed in other organisms.     

Transfer of training: Effects of successive pretraining of components in a dimension-abstracted oddity task

An experiment was conducted investigating the effects of training two components of a dimension-abstracted oddity problem, oddity responding and attention. All Ss were given a series of three problems. Type of oddity training was manipulated on Problem 1, and Ss learned (1) an oddity response in a problem whose format was the same as the final transfer problem, (2) an oddity response in a problem whose format was different from the final problem, or (3) no oddity response. Attention training was manipulated on Problem 2. The results indicated that Ss given same format oddity training and appropriate attention training were able to integrate that information on the final problem, that the probability of transferring the oddity response learned on Problem 1 to Problem 3 depended upon the similarity of the task formats, and that the two trained components tended to combine in an interactive fashion.     

Stimulus interaction and transfer of discrimination in children

Two discrimination learning experiments on third- and fourth-grade children are reported. In Expt I, Ss received either simultaneous or successive pretraining followed by simultaneous or successive transfer tasks in which an irrelevant dimension was varied either between or within settings. With the incorporation of a response-produced cues proposal, a modified version of the Hull-Spence theory (C. C. Spiker, Psychological Review, 1970, 77, 496–515) accounted for 81.2% of the variance of the group means and provided a satisfactory method of both quantifying and manipulating the mechanisms of associative transfer, even when the dimensions of pretraining and transfer differed. Because simultaneously pretrained Ss performed more as was expected than did successively pretrained Ss, it was hypothesized that Ss have an extraexperimental tendency to respond to dimensions varying within settings. This hypothesis was supported by the results of Expt II, in which an extensive shaping procedure was used for simultaneous, successive or control pretraining tasks followed by a successive transfer task in which the arrangement of two irrelevant dimensions was manipulated.