Accurately measuring nonconscious processing using a generative Bayesian framework.

Goldstein, Ariel; Sklar, Asael Y.; Siegelman, Noam

doi:10.1037/cns0000316

Cited by 7 publications

(18 citation statements)

References 79 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Statistical tools can address this and other issues that arise from the use of awareness measures that are less than perfectly reliable. Employing tools like the one we briefly described at the end of Section 2 as well as more advanced ones (e.g., Goldstein et al, 2021) would allow the field to advance with greater confidence in the veracity of results. This, we hope, will lead to more cumulative and accurate science and therefore to faster progress in what is one of the most intriguing questions in psychology.…”

Section: Discussionmentioning

confidence: 99%

Regression to the Mean Does Not Explain Away Nonconscious Processing

Sklar

Goldstein

Hassin

2021

Experimental Psychology

Self Cite

View full text Add to dashboard Cite

Abstract. In studies that use subliminal presentations, participants may become aware of stimuli that are intended to remain subliminal. A common solution to this problem is to analyze the results of the group of participants for whom the stimuli remained subliminal. A recent article ( Shanks, 2017 ) argued that this method leads to a regression to the mean artifact, which may account for many of the observed effects. However, conceptual and statistical characteristics of the original publication lead to overestimation of the influence of the artifact. Using simulations, we demonstrate that this overestimation leads to the mistaken conclusion that regression to the mean accounts for nonconscious effects. We conclude by briefly outlining a new description of the influence of the artifact and how it should be statistically addressed.

show abstract

Section: Discussionmentioning

confidence: 99%

Regression to the Mean Does Not Explain Away Nonconscious Processing

Sklar

Goldstein

Hassin

2021

Experimental Psychology

Self Cite

View full text Add to dashboard Cite

show abstract

“…Unfortunately, further studies showed that this method overestimates unconscious effects due to the underestimation of the regression slope (Dosher, 1998;Sand & Nilsson, 2016). In response to this problem, Goldstein et al (2022) recently developed a Bayesian generative model solution for estimating the intercept of the regression model while accounting for the measurement error.…”

Section: Introductionmentioning

confidence: 99%

“…To address these questions, in the present study we aim to explore unconscious visual processing through an integrative paradigm that includes objective and subjective measures of awareness collected both during the main task (online) and in separate blocks (offline). In addition, we will examine unconscious processing through the different classical analysis strategies associated with various combinations of awareness measures and research paradigms (e.g., dissociation paradigm, trial-wise post-hoc selection), together with some new approaches recently proposed by different authors and based on models derived from Bayesian analysis and General recognition theory (Goldstein et al, 2022;Pournaghdali et al, 2023).…”

Section: Introductionmentioning

confidence: 99%

In search of an integrative method to study unconscious processing: an application of Bayesian and General Recognition Theory models to the processing of hierarchical patterns in the absence of awareness.

Prieto,

Montoro,

Jimenez

et al. 2024

Preprint

View full text Add to dashboard Cite

The dissociation between conscious and unconscious perception is one of the most controversial issues in the study of consciousness. Some authors argue in favor of the existence of high-level unconscious processing, while others claim that most of the results obtained can be explained by a mixture of methodological issues and erroneous (or at least questionable) assumptions about what can be considered non-conscious. Particularly, one of the most debated topics involves the way in which the awareness of the stimuli is assessed, with a wide variety of measures and associated methods being proposed throughout the history of the field. To address this question, in the present study an integrative paradigm was devised to assess the extent to which masked hierarchical stimuli can be processed in the absence of awareness. We combined a primming task where participants were presented with Navon-like hierarchical masked stimuli, with the use of subjective and objective awareness measures collected either in a separate block (offline) or trial-by-trial during the main task (online). The unconscious processing of the masked primes was then evaluated through two different novel model-based methods. A Bayesian modeling approach and a General Recognition Theory modeling approach. Crucially, these two approaches assess unconscious processing through different theoretical assumptions and employing different awareness measures. The results showed that although there is a high correlation between awareness measures, the use of alternative approaches under different theoretical assumptions to analyze unconscious processing, leads to somewhat diverging conclusions about the extent of the unconscious processing of the masked primes.

show abstract

“…It is conceptually similar to selecting on the basis of a continuous variable, such as all d' values less than zero, and thereby ending up with a biased estimate of the mean d' of that sub-sample (for example, an estimate of less than zero for a true mean of above zero). But the suggested solutions for the latter problem (Goldstein et al, 2022;Kelley, 1947;Leganes-Fonteneau et al, 2021;4 Shanks, 2017;Yaron et al, 2023) either do not directly apply to the current one, or have not been shown to apply.…”

mentioning

confidence: 99%

How to deal with regression to the mean when selecting out conscious trials in order to analyse unconscious trials

Dienes

2024

Preprint

View full text Add to dashboard Cite

In implicit cognition research generally, one standard strategy is to measure the conscious status of knowledge on each trial (e.g. with confidence, structural knowledge attributions, visual clarity ratings) and then sub-select the trials where the knowledge is measured to be unconscious. If the accuracy is above chance on the latter trials that is taken to be evidence for unconscious knowledge. David Shanks (2017) has pointed out the problem of regression to the mean when trials (or people) are sub-selected based on a variable: Because of the ubiquitous possibility of error in measurement, the estimated value of the variable for the sub-set selected based on pre-selection scores will be biased. Thus, for example, trials selected to be based on unconscious knowledge will actually sometimes be based on conscious knowledge. Does this critique undermine the separate analyses of categories obtained every trial, such as structural knowledge attributions in implicit learning research, or confidence or PAS ratings in subliminal perception research? I show how to quantify the maximum effect that could be produced by regression to the mean in a given situation, how it may be a problem but it may also be so small as to be meaningless, and how to deal with it when it is of a moderate size, using Bayes factors with an interval null hypothesis.

show abstract

Accurately measuring nonconscious processing using a generative Bayesian framework.

Cited by 7 publications

References 79 publications

Regression to the Mean Does Not Explain Away Nonconscious Processing

Regression to the Mean Does Not Explain Away Nonconscious Processing

In search of an integrative method to study unconscious processing: an application of Bayesian and General Recognition Theory models to the processing of hierarchical patterns in the absence of awareness.

How to deal with regression to the mean when selecting out conscious trials in order to analyse unconscious trials

Contact Info

Product

Resources

About