共查询到20条相似文献，搜索用时 335 毫秒
1.
In Experiment 1, it was shown that generalization testing following successive discrimination training between two closely spaced wavelengths results in a sharp gradient with a peak of responding shifted from S+ so as to be further removed from S?. Testing after a 24-h delay resulted in a flatter gradient with greater peak (and area) shift. A 5-min pretest exposure to S+, reinforced or unreinforced, or to S? (unreinforced) reinstated immediate test performance; free reinforcement with no discriminative stimulus present had no such effect. Experiment 2 replicated the flattening of generalization gradients and enhanced peak shift in delayed testing. Free feeding in a pretest treatment with a distinctive food uniquely associated with the wavelength discrimination problem failed to reinstate immediate test performance. Experiment 3 tested the hypothesis that free feeding failed as a reactivation treatment because it did not engender keypecking. Subjects were trained to peck a vertical line stimulus before being given wavelength discrimination training. Again, the enhanced peak shift and greater flattening with delayed wavelength generalization testing was found. A pretest exposure to the vertical line stimulus elicited pecking but had no effect on subsequent wavelength generalization. Thus, only a reactivation treatment that included one of the discrimination training stimuli was effective in producing delayed test performance comparable to that obtained in an immediate test. 相似文献
2.
3.
The effects of test stimulus range on generalization gradients in humans were assessed for discriminations between faces that
varied in brightness, faces that varied in orientation in the picture plane, and morphed faces. In Experiment 1, significant
range effects, predicted by adaptation level theory, occurred when faces varied along the brightness or orientation dimension,
but not for morphed faces. Changing the difficulty of discrimination of both training and test stimuli for Experiment 2 produced
range effects in morphed faces. Experiment 3 explored training and testing stimulus factors as determinants of range effects
in morphed faces. The results suggest that sufficiently biased testing ranges create shifts in response distributions (generalization
gradients), and this may be amplified by using relatively difficult discriminations between training stimuli. 相似文献
4.
In Experiment 1, 12 pigeons were given eight sessions of VI single stimulus training with a color in a particular context followed by eight sessions of similar training with a line angle in another context. On the next day, half of the subjects were tested for wavelength and angularity generalization in each of the two contexts, a procedure that was thus consistent with training for one dimension and inconsistent for the other. The subjects made significantly more responses to each training stimulus under the consistent context condition, but there was no difference in absolute or relative generalization slopes. In Experiment 2, 12 pigeons were trained as in Experiment 1, but during generalization testing they were exposed to both contexts sequentially. Under the consistent context condition, the subjects responded more to the two training stimuli and yielded sharper absolute and relative wavelength generalization gradients: Under the inconsistent context condition, responding to the training wavelength was substantially disrupted. Thus, under appropriate testing conditions, contextual control over both the amount and the selectivity of responding can be demonstrated. 相似文献
5.
Learning to discriminate stimuli can alter how one distinguishes related stimuli. For instance, training an individual to
differentiate between two stimuli along a single dimension can alter how that individual generalizes learned responses. In
this study, we examined the persistence of shifts in generalization gradients after training with sounds. University students
were trained to differentiate two sounds that varied along a complex acoustic dimension. The students were subsequently tested
on their ability to recognize a sound that they had experienced during training when it was presented among several novel
sounds varying along this same dimension. Peak shift was observed in Experiment 1, in which generalization tests immediately
followed training, and in Experiment 2, in which the tests were delayed by 24 h. These findings further support the universality
of generalization processes across species, modalities, and levels of stimulus complexity. They also raise new questions about
the mechanisms underlying learning-related shifts in generalization gradients. The sound stimuli from this study are available
as .wav files from http://lb.psychonomic-journals.org/content/supplemental. 相似文献
6.
Four experiments are reported in which pigeons first learned one wavelength discrimination (green S+, yellow S?) and then the reversal; finally, after various delays, they were tested for wavelength generalization in extinction. In Experiment 1, the two problems were learned in different contexts; testing in Context 1 produced maximal responding to green in only half of the subjects, even when testing was delayed 30 days. In Experiment 2, testing of the subjects repeatedly in both contexts showed good control by each context after a 30-day delay. In Experiment 3, both problems were learned in the same context, and all gradients showed recency, peaking at yellow, even after 30 days. In Experiment 4, the subjects learned a series of reversals in the same context, terminating in yellow, S+, green, S?, and their gradients peaked at yellow, even after a 30-day delay. In Experiments 3 and 4, the gradients became flatter with increasing delays, and they were flatter in Experiment 4 (after three reversals) than in Experiment 3 (after one reversal). The location of the peak was not affected by delay, but only by testing in a context that had been uniquely associated with Problem 1 (Experiments 1 and 2). It is proposed that the location of gradient peaks indicates what is being remembered, whereas the slope of the obtained gradients indicates how well the target memory has been retrieved. 相似文献
7.
In Experiment 1, six groups of pigeons (n=8) were tested for wavelength generalization either immediately or 24 h after learning a successive discrimination, with 550 nm reinforced and a black vertical line extinguished. The groups differed in the stimulus present during single stimulus pretraining, which was 550 nm (pretrain S+), the vertical Une (pretrain S?), or a neutral dim white light (pretrain Sn), respectively. The three immediate generalization gradients were steep and indistinguishable, reflecting only the immediately preceding discrimination training condition. The three delay gradients were flatter, with the flattening particularly marked in the pretrain S? group. This was interpreted as proactive interference (PI) resulting from the memory that both the 550-nm and the line stimuli had previously been reinforced. In Experiment 2, two (TD) groups of pigeons (n=16) were given single stimulus training with a 555-nm keylight followed by eight sessions of discrimination training with two line angles, then one session of non-differential (ND) training with the same two lines, and then a wavelength generalization test either immediately or after a 24-h delay. Two other (hold) groups (n=16) received similar training, except for the TD Une angle training sessions, in these hold groups, the wavelength gradient was flatter in a delayed test; in the TD groups it was steeper, indicating PI from the prior TD training. These two experiments suggest that the “attentional sets,” which purportedly result from TD and ND training, may fruitfully be viewed as target memories subject to the principles of interference theory. 相似文献
8.
Pigeons trained monocularly and tested with the trained eye along the angularity dimension were found to show enhanced area shifts when compared to pigeons trained and tested binocularly. Pigeons trained monocularly and tested with the untrained eye showed reversed peak shifts in the direction of S?. These findings are consistent with the hypothesis that the mirror image of S? in angularity gradients functions to reduce the likelihood of the peak shift and area shift along the angularity dimension in pigeons, and question the use of the angularity dimension for intradimensional discrimination training and generalization testing. 相似文献
9.
Perceptron models have been used extensively to model perceptual learning and the effects of discrimination training on generalization, as well as to explore natural classification mechanisms. Here, we assess the ability of existing models to account for the time course of generalization shifts that occur when individuals learn to distinguish sounds. A set of simulations demonstrates that commonly used single-layer and multilayer perceptron networks do not predict transitory shifts in generalization over the course of training but that such dynamics can be accounted for when the output functions of these networks are modified to mimic the properties of cortical tuning curves. The simulations further suggest that prudent selection of stimuli and training criteria can allow for more precise predictions of learning-related shifts in generalization gradients in behavioral experiments. In particular, the simulations predict that individuals will show maximal peak shift after different numbers of trials, that easier contrasts will lead to slower development of shifted peaks, and that whether generalization shifts persist or dissipate will depend on which stimulus dimensions individuals use to distinguish stimuli and how those dimensions are neurally encoded. 相似文献
10.
Pigeons’ responses on an operant key were reinforced according to either multiple variable-interval variable-interval or multiple variable-interval extinction schedules. The multiple-schedule components were signaled by line-tilt stimuli on a second key (signal key). Signal-key responses never produced reinforcement, and operant-key responses were not reinforced if they followed within 1 sec of a signal-key response. Behavioral contrast was not observed on the operant key, although there was a small, but reliable, increase in signal-key responding in the variable-interval component of the multiple variable-interval extinction condition. Generalization tests were interspersed between sessions of multiple variable-interval extinction training. Generalization gradients along the line-tilt dimension exhibited peak shift for both operant-key and signal-key responding following intradimensional (line tilt) discrimination training. Line-tilt generalization gradients following interdimensional discrimination did not exhibit peak shift. Gradients following intradimensional discrimination were sharper than gradients following interdimensional discrimination for both operant-key and signal-key responding. It was concluded that dimensional stimulus control of topographically tagged responding maintained by the stimulusreinforcer relation parallels that maintained by the response-reinforcer relation. 相似文献
11.
David R. Thomas C. F. Hickis Raymond L. Jackson Robert J. Newlin 《Learning & behavior》1979,7(3):301-308
In Experiment 1, two groups (n = 10) of pigeons received 17 sessions of TD (true discrimination) or ND (nondifferential) training with line angles. Seventeen sessions of SS (single stimulus) training with a wavelength preceded this training and two followed it. Subsequent wavelength generalization testing in extinction revealed a sharper TD than ND gradient. This slope difference was evident from the very first test stimulus presentation and remained stable throughout testing. As a consequence of substantial overtraining, there was no reduction of response strength and no sharpening of generalization during testing for either group. In Experiment 2, two groups (n = 16) of pigeons received 10 sessions of TD or PD (pseudodiscrimination) training with line angles, followed by four sessions of SS training with a single wavelength. During this training and in subsequent wavelength generalization testing in extinction, brief blackouts separated stimulus presentations. Again, the TD group yielded the sharper gradient. Although responding weakened and the gradients sharpened during the test, these effects were comparable in the two groups. Furthermore, gradients based on the percentage of trials with at least one response showed the same TD-PD slope difference. This finding indicates that differential control over responding by response-produced feedback is inadequate to account for the TD-PD difference in generalization slope. Both experiments indicate that a purported difference in resistance to extinction is also an inadequate explanation. 相似文献
12.
Robert A. Hancock 《Learning & behavior》1982,10(1):46-54
Egger and Miller (1962) hypothesized that the conditioned reinforcing value of stimuli depends on their information value. Egger and Miller and others have tested this hypothesis by comparing the conditioned reinforcing value of S1 and S2 following S1-S2-reward training. However, none of these experiments have controlled for differential generalization of conditioned reinforcement value from training to comparison tests. That is, the S1 cue pattern during the conditioned reinforcement tests has been very similar to the S1 cue pattern of training, while the training and test S2 cue patterns have been quite dissimilar. In Experiment 1, pigeons in a procedure unconfounded by differential generalization produced S2 reliably more frequently than S1, and pigeons in a confounded procedure produced S1 somewhat more frequently than S2. A significant groups × stimuli interaction was attributed to differential stimulus generalization from training to test for S1 and S2 in the confounded condition. In Experiment 2, pigeons in an unconfounded procedure again produced S2 reliably more frequently under a different testing procedure. The results are interpreted as demonstrating that, following S1-S2-food training trials, S2 is the more effective conditioned reinforcer in unconfounded conditions. A reconceptualization of the information hypothesis is shown to be consistent with these results. 相似文献
13.
In Stage 1, four groups of pigeons were given true discrimination (TD) and four groups were given pseudodiscrimination (PD) training along one of two dimensions orthogonal to wavelength. In Stage 2, all groups received single stimulus (SS) training with a wavelength cue compounded with the former S+ for four of the groups (two TD and Two PD) and with a novel stimulus for the other four (two TD and two PD). For all groups, the SS training was followed by a wavelength generalization test in order to assess the control acquired by the (incidental) wavelength cue during Stage 2. The presence of the former S+ should have blocked the acquisition of control by the wavelength cue leading to flat generalization gradients in the appropriate TD groups. To the contrary, however, these groups showed the sharpest wavelength gradients, i.e., rather than blocking, potentiation was found. One possible interpretation is that the superimposition of the previous S+ served as a reminder of previous TD training, thereby enhancing the transfer of an attentive set from Stage 1 to Stage 2. An alternative conditioning interpretation, suggested in a 1975 paper by Mackintosh, can also be extended to encompass these results. 相似文献
14.
David R. Thomas Alan R. McKelvie Michael Ranney Thomas B. Moye 《Learning & behavior》1981,9(4):581-586
In Experiment 1, three (experimental) groups of pigeons (n=8) acquired a successive wavelength discrimination in Phase 1 and a reversal in Phase 2; then, after a 24-h delay, they received a wavelength generalization test in extinction (Phase 3). For one group (“Context Same”), the same context was present throughout; for both others, a different context was used for Phase 1 and Phase 2. One group received the generalization test in the presence of Context 1, the other in Context 2. The Context Same and Context 2 experimental groups showed “recency,” with all gradients peaking at the reversal S+ value. The Context 1 group yielded several different response patterns but never showed recency, thus revealing context-generated proactive interference. In Experiment 2, eight subjects learned the original discrimination and its reversal in different contexts, and each bird was tested alternately (within a session) in both contexts. Under this condition, the test contexts were effective retrieval cues. In every case, the gradients obtained in each context peaked sharply at the appropriate S+ value. These experiments indicate that conflicting memories may be stored along with their associated contexts such that they can be retrieved by an appropriate manipulation of contextual cues at the time of retention testing. 相似文献
15.
Young adult subjects were trained in a discrimination involving three pure tones varying in pitch, with two positive stimuli and a negative stimulus located midway between the two. Subjects in the control conditions were trained only with the two positive stimuli. Generalization gradients were bimodal in all cases, but for the control subjects, the modes were at the positive stimulus values, while postdiscrimination gradients were displaced away from the negative stimulus toward the extremes of the continuum. Since, with this procedure, the training and test adaptation levels were the same, the observed displacements are difficult to explain in terms of an adaptation-level account of peak shifts. The observed shifts are more consistent with the conditioning-extinction model which assumes interactions between inhibitory and excitatory gradients. However, overall response rates did not reflect the action of summation processes. 相似文献
16.
Four naive pigeons were given six generalization tests in extinction after periods of pretraining in which S+ appeared with food reinforcement and S? appeared in extinction. An analysis of sequential effects among presentations of test stimuli showed that the overall gradient was influenced differently by stimuli at the extremes of the continuum of test stimuli and by S+ and adjacent stimuli. Gradients consisting of responding in each stimulus when it was preceded by an extreme stimulus tended to peak at S+, while gradients produced when each stimulus was preceded by S+ or an adjacent stimulus tended to show a peak shift. This was true whether the overall gradient showed a peak shift or not. Two naive subjects were added and four additional tests were given after pretraining in which unequal frequencies of reinforcement accompanied both S+ and S?. Results of all 10 tests show that sequential effects occur during generalization testing in extinction and that these “local dimensional effects” are unlike local contrast. These stimulus-specific sequential effects may greatly influence overall gradient form. 相似文献
17.
James V. Couch 《Learning & behavior》1975,3(4):347-358
Two experiments examined the presumed relationship between behavioral contrast and inhibitory stimulus control. In Experiment I, pigeons were exposed to mult VI 1-min VI 1-min or mult VI 5-min VI 5-min during baseline training prior to mult VI 1-min VI 5-min discrimination training. Half of the subjects received a timeout (TO) component during baseline in order to reduce the degree of contrast during discrimination training. Only 3 of 8 subjects receiving the TO showed contrast while all other subjects showed various degrees of contrast. Postdiscrimination generalization gradients indicated excitatory rather than inhibitory control by the stimulus associated with the VI 5-min schedule. During baseline training in Experiment II, responding to all the generalization stimuli was reinforced. In addition, some subjects received the TO stimulus. The subjects were next exposed to mult VI 1-min EXT, mult VI 1-min VI 5-min, or just the VI 5-min component. Generalization gradients indicated inhibitory control by the stimulus associated with EXT or VI 5-min for 19 of 20 subjects even though some subjects did not show contrast. These results question the presumed relationship between behavioral contrast and inhibitory stimulus control. 相似文献
18.
The present research investigated the effects of physical context change and perceptual learning on generalization. In a video game, participants learned to suppress their mouse-clicking behavior in the presence of one stimulus (AX). Generalization was observed between the AX stimulus and another stimulus (BX) that was designed to be similar. When testing was conducted in a context different from that in which AX was used in training, responding to AX was attenuated, and responding to BX was enhanced. That is, the generalization gradient flattened. The latter effect was only evident in groups for which generalization had been reduced through a preexposure manipulation believed to produce perceptual learning. Experiment 2 demonstrated that the increase in generalization observed in the first experiment was due to the context change between the preexposure and test rather than to a change between the conditioning and test contexts. Implications for flattening generalization gradients and mechanisms of perceptual learning are discussed. 相似文献
19.
Rats received three-trial series on a T-maze consisting of extended visually distinct left-black and right-striped side runways.
During the first phase of training, when allowed to select baited runways within these series, they predominantly alternated
their choices. During the second phase, rats received forced-choice serial pattern training of series consisting of two rewarded
(R) trials and one nonrewarded (N) trial in two fixed orders, RRN and RNR. In Experiment 1, the rats in the runway shift rule
group always received the second R trial when forced down a runway opposite that on the preceding trial in the series and
the N trial when forced down the same runway. The rats in the runway stay rule group always received the second R trial when
forced down the same runway and the N trial when forced down the opposite runway. In Experiment 2, each rat was conditionally
trained with both runway outcome rules as determined by the central alley lighting and the type of food in the side alleys.
The rats took longer to reduce their running speed on the N trial within each sequence under the runway stay rule than under
the runway shift rule. They also took longer to acquire serial pattern responding for the RNR than for the RRN series only
under the runway stay rule condition. When subsequently reexposed to series of free-choice trials on the final phase, rats
maintained spontaneous alternating choice patterns under the runway shift rule conditions but either seldom alternated their
choices (Experiment 1) or greatly reduced choice alternations (Experiment 2) under the runway stay rule condition. We discussed
these effects in terms of rats’ natural foraging strategies and as a factor that interacts with other within- and between-series
variables that affect serial pattern behavior. 相似文献
20.
Steven J. Haggbloom 《Learning & behavior》1983,11(3):367-372
In two experiments, rats were trained on a successive go/no-go discrimination problem in the runway in which the positive (S+) and negative (S?) discriminanda were differentiated by the presence or absence of a distinctive feature. The feature in Experiment 1 was a series of flashing lights over the runway. In Experiment 2, the feature was a pretrial reinforcement (Phase 1), or pretrial reinforcement versus pretrial nonreinforcement (Phase 2). The feature signaled S+ trials in feature-positive (FP) groups and S? trials in feature-negative (FN) groups. The original discrimination was reversed in Phase 2 of both experiments. With the exception of the pretrial nonreinforcement groups in Experiment 2, there was an asymmetry in discrimination learning in both phases of both experiments favoring superior discrimination learning by FN subjects over FP subjects, a feature-negative effect. Implications of the results for an information processing account of asymmetries in learning feature discriminations are discussed. 相似文献