Revisiting peak shift on an artificial dimension: Effects of stimulus variability on generalisation

Livesey, Evan J.; McLaren, Ian A.

doi:10.1177/1747021817739832

Cited by 11 publications

(12 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

Section: A New Approach: Curve-fitting and Parameter Estimation Usingmentioning

confidence: 99%

“…Livesey and McLaren (2019) have recently demonstrated how Gaussian curve-fitting can be successfully applied to quantify the degree of peak shift. Livesey and McLaren (2019) fitted Gaussian functions to individual gradients using maximum likelihood estimation and plotted the best-fitting estimates for each participants’ peak location (the mean of the Gaussian function). This allowed them to quantify the degree of peak shift and also to directly compare the size of the peak shift effect between experimental conditions, something that is not possible with the more common method of peak shift analysis described above.…”

Section: A New Approach: Curve-fitting and Parameter Estimation Usingmentioning

confidence: 99%

“…This allowed them to quantify the degree of peak shift and also to directly compare the size of the peak shift effect between experimental conditions, something that is not possible with the more common method of peak shift analysis described above. However, as the technique used by Livesey and McLaren (2019) reduced each participants’ gradient to a single data point (the best-fitting estimate of the Gaussian mean), there was substantial information loss. Therefore, large sample sizes may be needed to observe clear group differences in the distribution of the estimates.…”

Section: A New Approach: Curve-fitting and Parameter Estimation Usingmentioning

confidence: 99%

See 2 more Smart Citations

Modelling generalisation gradients as augmented Gaussian functions

Lee

Mills

Hayes

et al. 2020

Quarterly Journal of Experimental Psychology

Self Cite

View full text Add to dashboard Cite

Section: A New Approach: Curve-fitting and Parameter Estimation Usingmentioning

confidence: 99%

Section: A New Approach: Curve-fitting and Parameter Estimation Usingmentioning

confidence: 99%

Section: A New Approach: Curve-fitting and Parameter Estimation Usingmentioning

confidence: 99%

See 1 more Smart Citation

Modelling generalisation gradients as augmented Gaussian functions

Lee

Mills

Hayes

et al. 2020

Quarterly Journal of Experimental Psychology

Self Cite

View full text Add to dashboard Cite

“…The authors argue that their results imply the involvement of rule-based processes in the generalisation of conditioned fear along simple stimulus dimensions. Peak shift continues to be the phenomenon of interest in the third paper in this Special Issue in which Livesey and McLaren (2019), like the authors in the previous paper, focus on the distinction between learning about features and learning about relations. Their paper reports a series of experiments inspired by Wills and Mackintosh's peak-shift experiments with people and pigeons and investigated the processes that are involved in the establishment of the gradient of generalisation following human category learning.…”

mentioning

confidence: 92%

The psychology of associative learning: Editorial

Haselgrove

McLaren

2019

Quarterly Journal of Experimental Psychology

View full text Add to dashboard Cite

The articles in this Special Issue have been chosen to cover the rather wide range of topics that fall under the heading "Associative Learning." As such, they represent one strand of the work of Nick Mackintosh, who passed away in February 2015, with other aspects of his research, notably his work on individual differences and IQ, only touched on in passing. Nevertheless, even though they do not encompass the full range of Nick's interests, they still convey something of the sheer scope of his contribution to our subject, and the way in which he was able to take a particular approach to psychology and apply it to almost any subdomain of that discipline. The papers brought together in this volume are testament to the fact that this approach still commands our attention today and still has an important role in psychology at this time. An obituary to Nick is published in Biographical Memoirs of Fellows of the Royal Society (Pearce, 2018).The first article in this Special Issue is the Inaugural Mackintosh Lecture given at the Associative Learning Symposium held at Gregynog Hall, Wales, Easter 2016. This lecture series was established in honour of Nick's contribution to psychology and in particular to associative learning. In it, McLaren et al. (2019) make the case for associative learning in humans, an issue that Nick had worked on with several of the authors previously. Using paradigms ranging from Pavlovian conditioning in humans to task switching and sequence production, they show that there are more than propositional processes at work controlling behaviour and provide evidence for an associative component as well. Where this approach differs from more traditional dual-process theories is in emphasising the way that associative and rule-based processes are different facets of the same learning system. It is not that there are two systems operating in parallel; instead, the view is that there is one system that amounts to propositional processes constructed from an associative substrate. Such a system may typically seem propositional in nature, but can be persuaded to reveal its associative underpinnings.The second article, by Ahmed and Lovibond (2019), explores the peak shift effect, studied in both pigeons and humans by Sarah Wills and Nick Mackintosh (1998). Peak shift refers to the observation that, following training of two stimuli that differ along points on a dimension as a CS+ and CS−, conditioned responding is strongest not to the CS+ but to a stimulus further along the dimension from the CS+, in the opposite direction to the CS−. Lovibond and Ahmed show, in two shock conditioning experiments with human participants, that this effect can be modulated by verbalisable rules. In both experiments, different gradients were observed when participants were divided on the basis of the rules they reported using during the task (linear and similarity). The authors argue that their results imply the involvement of rule-based processes in the generalisation of conditioned fear along simple stimulus dimensions. Peak s...

show abstract

“…On the other hand, if participants are using a relational rule, then categorization accuracy and ratings of typicality should diverge as the test stimuli become less similar to the trained stimuli, with a sharp decline in typicality ratings at the extreme end of the dimension. In summary, the current study aims to investigate whether peak shift can be found in the presence of a relational rule by disrupting rule application, rather than hindering rule formation ([ 22 , 31 ]) or assessing generalization prior to rule formation ([ 20 ]) as previous studies have done. In addition, by including typicality ratings, a monotonic gradient of generalization can be more strongly inferred to be the product of relational rule use.…”

Section: Introductionmentioning

confidence: 99%

Rule-based generalization and peak shift in the presence of simple relational rules

Lee

Livesey

2018

PLoS ONE

View full text Add to dashboard Cite

After discrimination learning between two stimuli that lie on a continuum, animals typically exhibit generalization on the basis of similarity to the physical features of the stimuli, often producing a peak-shifted gradient. However, post-discrimination generalization in humans usually resembles a monotonically increasing (e.g., linear) gradient that is better characterized as following a relational rule describing the difference between the stimuli. The current study tested whether rule-based generalization could be disrupted by reducing the applicability of a relational rule on test. We compared generalization following a difficult categorization task between a group who could use their rule consistently throughout test (Group Consistent), and a group who could only apply their rule effectively on 50% of test trials and thus could only use it inconsistently (Group Inconsistent). Across two experiments, a peak shift was found in the Inconsistent group and a monotonic gradient in the Consistent group. A post-hoc sequential analysis revealed that the Inconsistent group produced both peak-shifted and monotonic gradients as a function of whether or not the relevant rule was applicable on the previous trial. Reducing the applicability of a rule on test thus appeared to lead participants to revert to generalizing on the basis of similarity. Our results suggest that humans learn about the physical features of the stimuli alongside relational rules, and that rule- and similarity-based learning can interact in determining generalization.

show abstract

Revisiting peak shift on an artificial dimension: Effects of stimulus variability on generalisation

Cited by 11 publications

References 39 publications

Modelling generalisation gradients as augmented Gaussian functions

Modelling generalisation gradients as augmented Gaussian functions

The psychology of associative learning: Editorial

Rule-based generalization and peak shift in the presence of simple relational rules

Contact Info

Product

Resources

About