Negative evidence and inductive reasoning in generalization of associative learning.

Lee, Jessica C.; Lovibond, Peter F.; Hayes, Brett K.; Navarro, Daniel J.

doi:10.1037/xge0000496

Cited by 30 publications

(23 citation statements)

References 73 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the present work, we focus on heterogeneity in specifications within the BLC category in comparison to one prototypical TTP approach. All of the 18 studies, except one using ANSLAB (Rattel et al, 2019) used a custom-made script whereof three studies (Ahmed & Lovibond, 2019;Lee et al, 2018Lee et al, , 2019 explicitly reported computing SCL rather than SCR. Yet, we excluded one study that investigated SCL without a baseline correction (McGlade et al, 2019), and one study that did not report any detail on SCL quantification (Kurayama et al, 2019).…”

Section: Systematic Literature Searchmentioning

confidence: 99%

A data multiverse analysis investigating non-model based SCR quantification approaches

Sjouwerman¹,

Illius²,

Kuhn³

et al. 2021

Preprint

View full text Add to dashboard Cite

Data inevitably need to be processed, typically involving multiple decision nodes with decisions often being equally justifiable. Electrodermal signals are the most common outcome measure in fear conditioning research, but response quantification approaches vary strongly. It remains an open question whether different approaches result in convergent results. Using fear conditioning research as a case example, we identified that baseline-correction (BLC) and trough-to-peak (TTP) quantification are used most frequently in the literature. Furthermore, heterogeneity of specifications in BLC formulas was observed, i.e., within the pre-CS baseline window and the post-CS peak detection or mean detection window. Here we systematically scrutinize the robustness of results when applying different processing methods to one pre-existing dataset (N= 118). The study was pre-registered. We report high agreement between different BLC approaches for US and CS+ trials, but moderate to poor agreement for CS- trials. Furthermore, a specification curve of the main effect of CS discrimination during fear acquisition training from all potential and reasonable combinations of specifications (N=150) and a prototypical TTP approach indicates that resulting effect sizes are largely comparable. Crucially, however, we show that BLC approaches often misclassify the peak SCR - particularly for CS- trials, which leads to a stimulus-specific bias and challenges for post-processing and replicability. Lastly, we investigate how physiologically implausible (negative) skin conductance values in BLC appearing most frequently for CS- (CS- > CS+ > US), correspond to in TTP quantification. We discuss the results in terms of robustness and replicability and provide insights into challenges, opportunities, and implications.

show abstract

Section: Systematic Literature Searchmentioning

confidence: 99%

A data multiverse analysis investigating non-model based SCR quantification approaches

Sjouwerman¹,

Illius²,

Kuhn³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Despite sharing many important commonalities in the way they conceptualize learning and generalization, associative learning and inductive reasoning were studied as separate literatures. Recently, researchers have started to acknowledge the importance of incorporating inductive principles in the study of fear/aversive conditioning (Dunsmoor & Murphy, 2015;Dymond et al, 2015;Hayes & Heit, 2018;Lee, Hayes, & Lovibond, 2018;Lee, Lovibond, Hayes, & Navarro, 2019;Lei et al, 2019;Lei et al, 2019). It was proposed that aversive or traumatic experiences promote inductive generalization.…”

Section: Semantic Processing Of Subliminal Wordsmentioning

confidence: 99%

“…inductive generalization differs among different types of stimuli, such as fear versus nonfear generalization in clinical populations. This could help identify the mechanisms underlying overgeneralization observed in PTSD and other anxiety disorders (Lee et al, 2019).…”

Section: Semantic Processing Of Subliminal Wordsmentioning

confidence: 99%

Unconscious preparation: Effects of prime visibility on semantic generalization of task priming

Zher-Wen

2021

British J of Psychology

View full text Add to dashboard Cite

Studies found that subliminal primes can be associated with specific tasks to facilitate task performance, and such learning is highly adaptive and generalizable. Meanwhile, conditioning studies suggest that aversive/reward learning and generalization actually occur at the semantic level. The current study shows that prime-task associations can also be generalized to novel word/neighbour primes from the same semantic category, and this occurs without contingency awareness. Previous studies have counterintuitively suggested that both the learning of task priming and the semantic priming of word neighbours depend on the lack of visibility. Here, we show that semantic generalization indeed depends on reduced visibility, but cannot occur subliminally. The current study shows for the first time that semantic learning and generalization can occur without any emotional or motivational factors, and that semantic priming can occur for arbitrarylinked stimuli in a context completely devoid of semantics.

show abstract

“…The learner updates beliefs in these hypotheses as new evidence is observed, with certain hypotheses becoming stronger and others becoming weaker. As well as changes in generalization due to additional types, Bayesian models can account for a range of inductive phenomena such as the effects of premise diversity (Hayes, Navarro, Stephens, Ransom, & Dilevski, 2019), the impact of samples containing both positive and negative evidence (Lee, Lovibond, Hayes, & Navarro, 2019;Voorspoels, Navarro, Perfors, Ransom, & Storms, 2015), and generalization based on causal rather than categorical relations (Kemp & Tenenbaum, 2009).…”

Section: The Effect Of Adding Types In Property Inductionmentioning

confidence: 99%

Adding Types, But Not Tokens, Affects Property Induction

2020

Self Cite

View full text Add to dashboard Cite

The extent to which we generalize a novel property from a sample of familiar instances to novel instances depends on the sample composition. Previous property induction experiments have only used samples consisting of novel types (unique entities). Because real‐world evidence samples often contain redundant tokens (repetitions of the same entity), we studied the effects on property induction of adding types and tokens to an observed sample. In Experiments 1–3, we presented participants with a sample of birds or flowers known to have a novel property and probed whether this property generalized to novel items varying in similarity to the initial sample. Increasing the number of novel types (e.g., new birds with the target property) in a sample produced tightening, promoting property generalization to highly similar stimuli but decreasing generalization to less similar stimuli. On the other hand, increasing the number of tokens (e.g., repeated presentations of the same bird with the target property) had little effect on generalization. Experiment 4 showed that repeated tokens are encoded and can benefit recognition, but appear to be given little weight when inferring property generalization. We modified an existing Bayesian model of induction (Navarro, Dry, & Lee, 2012) to account for both the information added by new types and the discounting of information conveyed by tokens.

show abstract

Negative evidence and inductive reasoning in generalization of associative learning.

Cited by 30 publications

References 73 publications

A data multiverse analysis investigating non-model based SCR quantification approaches

A data multiverse analysis investigating non-model based SCR quantification approaches

Unconscious preparation: Effects of prime visibility on semantic generalization of task priming

Adding Types, But Not Tokens, Affects Property Induction

Contact Info

Product

Resources

About