Further analysis of the overtraining reversal effect.

Mackintosh, N. J.

doi:10.1037/h0026784

Cited by 79 publications

(60 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With regard to the animal learning literature, the results of certain studies indicate that stimuli that have previously been r experienced as reliable predictors will maintain a higher t alpha (and hence, will subsequently be learned about more rapidly) than will stimuli experienced as nonpredictive (Bennett et al, 2000;Dopson et al, in press;George & Pearce, 1999;Mackintosh, 1969;Mackintosh & Little, 1969). Such findings fit well with the theoretical stance taken by Mackintosh (1975;hereafter, referred to as the Mackintosh model d ; see also Kruschke, 2001;Sutherland & Mackintosh, 1971).…”

mentioning

confidence: 99%

Learned predictiveness effects following single-cue training in humans

Pelley

Turnbull

Reimers

et al. 2010

Learning & Behavior

View full text Add to dashboard Cite

126One of the primary aims of associative learning theory and the research on which it is based is to establish the factors that influence learning-the factors that determine why, under a certain set of circumstances, an organism might learn more about the consequences of Stimulus X than about those of Stimulus Y. Evidence from the field of animal conditioning suggests that one such factor is the p prior predictive history of a stimulus. In particular, previous experience of a cue (e.g., a tone) as being predictive or nonpredictive of an outcome event (e.g., food) will influence the rate of subsequent learning about that cue (e.g., Bennett, Wills, Oakeshott, & Mackintosh, 2000; Dopson, Esber, & Pearce, in press;Hall & Pearce, 1979, 1982 Haselgrove, Esber, Pearce, & Jones, in press;Lubow & Moore, 1959;Mackintosh & Little, 1969;Mackintosh & Turner, 1971;Wilson, Boumphrey, & Pearce, 1992; see Le Pelley, 2004, for a review). And consistent with the suggestion that a common associative-learning mechanism might underlie animal conditioning and human contingency learning (Dickinson, 2001), several recent studies have demonstrated an influence of prior predictiveness on the rate of learning about cues in human learning (e.g., Bonardi, Graham, Hall, & Mitchell, 2005;Griffiths & Le Pelley, 2009;Kruschke, 1996;Le Pelley & McLaren, 2003;Le Pelley, Schmidt-Hansen, Harris, Lunter, & Morris, 2010;Livesey & McLaren, 2007;Lochmann & Wills, 2003;Maes, Damen, & Eling, 2004). This rate of learning about the cue is often referred to as the cue's associability, or, perhaps more controversially, the attention paid to the cue. In this article, we remain uncommitted on exactly what aspect of cue processing is influenced by previous experience of predictiveness-specifically, whether such differences in processing truly reflect differences in attention (see Le Pelley, in press-a). Hence, from this point on, we will use the theoretically neutral label alpha to refer to that property of cue processing that is influenced by the prior predictiveness of the cue.The problem posed by the literature cited above is that f the specific way in which the learned predictiveness of h stimuli influences alpha seems very inconsistent. With regard to the animal learning literature, the results of certain studies indicate that stimuli that have previously been r experienced as reliable predictors will maintain a higher t alpha (and hence, will subsequently be learned about more rapidly) than will stimuli experienced as nonpredictive (Bennett et al., 2000; Dopson et al., in press;George & Pearce, 1999;Mackintosh, 1969;Mackintosh & Little, 1969). Such findings fit well with the theoretical stance taken by Mackintosh (1975; hereafter, referred Kruschke, 2001;Sutherland & Mackintosh, 1971). Other studies, however, seem to demonstrate exactly the opposite, showing faster learning about stimuli that have previously been unreliable predictors, as compared with those experienced as consistently predictive (e.g., Haselgrove et al., in press;Swan & Pearce, 198...

show abstract

mentioning

confidence: 99%

Learned predictiveness effects following single-cue training in humans

Pelley

Turnbull

Reimers

et al. 2010

Learning & Behavior

View full text Add to dashboard Cite

show abstract

“…This was the overtraining reversal effect, the paradoxical observation that rats trained on a simultaneous discrimination learned the reversal of this task more readily when they had been given overtraining on the original (e.g., Mackintosh, 1969). The interpretation offered by analyzer theory was that overtraining was so effective in strengthening the relevant analyzer that the beneficial effects of sustained attention to the relevant dimension were able to outweigh any potential disadvantages of such training.…”

Section: Analyzer Theorymentioning

confidence: 99%

Learned Changes in Stimulus Representations (A Personal History)

Hall

2007

Span. J. Psychol.

View full text Add to dashboard Cite

Almost 40 years ago I began what turned out to be a programme of research on the way in which experience can change the effectiveness of the events used as stimuli in standard associative learning procedures. In this personal history I will describe my early (failed) attempts to find evidence for the acquired distinctiveness of cues, and my conclusion that experience tends to reduce, not enhance the associability of stimuli. I then go on to describe my attempts to square this conclusion with the stubborn empirical fact that, in some circumstances, pretraining with (or preexposure to) stimuli, can facilitate subsequent discrimination between them. I describe experiments (conducted mostly with rats as the subjects) showing how some of these effects can be explained in associative terms. Others, however, seemed to demand an explanation in terms of a new learning process that modulates the effective salience of stimuli. I go on to describe attempts to specify the nature of this process, and (bringing the story up to date) to describe recent experiments investigating the effects of salience modulation in human perceptual learning. Keywords: associative learning, stimulus discrimination, pretraining, stimulis salience, perceptual learningHace casi 40 años, empecé lo que con el tiempo se convirtió en un programa de investigación sobre la forma en que la experiencia puede cambiar la efectividad de los eventos empleados como estímulos en procedimientos típicos de aprendizaje asociativo. En esta historia personal, describiré mis primeros (fallidos) intentos de demostrar la distintividad adquirida de las claves, y mi conclusión de que la experiencia tiende a reducir, en vez de a facilitar, la asociabilidad de los estímulos. Después paso a describir mis intentos de hacer compatible esta conclusión con el innegable hecho empírico de que, en algunas circunstancias, el pre-entrenamiento con (o la pre-exposición a) los estímulos puede facilitar la posterior discriminación entre ellos. Describo los experimentos (llevados a cabo con ratas como sujetos) que muestran cómo algunos de estos efectos pueden explicarse en términos asociativos. Sin embargo, otros parecen exigir una explicación en términos de un nuevo proceso de aprendizaje que modula la saliencia efectiva de los estímulos. Paso a describir los intentos de especificar la naturaleza de este proceso y (llegando al momento actual) a describir los experimentos recientes que investigan los efectos de modulación de la saliencia en el aprendizaje perceptual humano. Palabras clave: aprendizaje asociativo, discriminación estimular, pre-entrenamiento, saliencia estimular, aprendizaje perceptual

show abstract

“…Since attentional models assume that cues of greater salience are more reliably attended, less can be gained through overtraining with such cues than through overtraining with cues of lesser salience. This argument has been used by Mackintosh (1969) to explain why the overlearning reversal effect more frequently occurs when a difficult discrimination is employed than when an easy discrimination is employed. Within the present design, this argument leads to the prediction that overtraining with cues of greater salience will interfere less with shift learning than will overtraining with cues of lesser salience.…”

Section: Comparison Of the Nonovertrained Groupsmentioning

confidence: 99%

The effects of cue salience and overtraining on extradimensional shift learning

Basden

1970

Psychon Sci

View full text Add to dashboard Cite

Rats were first trained to criterion with either more disparate (MD) or less disparate (LD) brightness cues relevant and orientation cues irrelevant; half were then shifted to the orientation discrimination and half were overtrained prior to the shift. For nonovertrained Ss, there was a trend toward more rapid EDS learning when previous training was with the MD brightness cues. Overtraining significantly facilitated shift learning when initial training was with the LD brightness cues but slightly retarded shift learning when initial training was with the MD brightness cues, producing a significant interaction. Cue salience was concluded to be a major determiner of relative speed of EDS learning.The present study investigated the effects of the level of initial training and of the relative salience of the cues of two dimensions upon learning rates in an extradimensional shift (EDS) task. In previous investigations, overtraining has sometimes retarded (e.g., Mackintosh, 1962) and sometimes facilitated EDS learning (e.g., Sutherland & Andelman, 1969), but more frequently has had no effect (e.g., Tighe, Brown, & Youngs, 1965). Although some of the inconsistency in previous results can be attributed to variations in design, even when similar designs are considered, some inconsistency remains. For example, with the design employed in the present study, i.e., the dimension relevant in each stage is a variable irrelevant dimension in the other, Siegel (1967), Uhl (1964), and Brookshire, Warren, & Ball (1961) found that overtraining did not affect the speed of EDS learning with rats. Brookshire et al (1961) in a replication with chicks and Schade & Bitterman (1965) with pigeons found that overtraining significantly retarded EDS learning but in each case in only one of the two shift conditions employed in the counterbalanced designs. Employing only one task order, Mackintosh (1963) also found that overtraining retarded EDS learning. These facts are consistent with the hypothesis that differential cue salience is responsible for the residual inconsistency.This hypothesis was tested by training or overtraining separate groups on easy and difficult discriminations, respectively, and then shifting all groups to the 'same' discrimination (i.e., one having the same relevant cues for all groups). This provided a comparable shift-learning baseline against *This report is based on a portion of a doctoral dissertation submitted to the University of California, Santa Barbara. tNow on NIHD Postdoctoral Fellowship at the University of California, Berkeley.Psychon. Sci., 1970, Vol. 20 (1) which the influence of differential cue salience in initial learning could be measured. The cues relevant in initial learning were two sets of differential brightness cues, a more disparate (MD) set and a less disparate (LD) set; these were compounded with a single set of stripe orientation cues relevant only during shift learning. A previous study (Basden, 1969) employing techniques similar to those of Kendler, Basden, & Bruckner (1970) had establi...

show abstract

Further analysis of the overtraining reversal effect.

Cited by 79 publications

References 43 publications

Learned predictiveness effects following single-cue training in humans

Learned predictiveness effects following single-cue training in humans

Learned Changes in Stimulus Representations (A Personal History)

The effects of cue salience and overtraining on extradimensional shift learning

Contact Info

Product

Resources

About