The Representation of Simple Ensemble Visual Features Outside the Focus of Attention

Alvarez, George A.; Oliva, Aude

doi:10.1111/j.1467-9280.2008.02098.x

Cited by 350 publications

(421 citation statements)

References 21 publications

Supporting

Mentioning

382

Contrasting

Unclassified

Order By: Relevance

“…Cohen and Dennett emphasized gist and ensemble representations as well as generic representations [29]. One illustration of gist/ensemble representations was provided by Alvarez [51], who showed that when subjects track 4 moving disks they also are aware of the 'center of mass' of 4 distractor disks. For simplicity, I will lump gist-like and generic perception together under the heading of generic representation.…”

Section: Generic Consciousness Combined With Unconscious Iconic Memorymentioning

confidence: 99%

Perceptual consciousness overflows cognitive access

Block

2011

Trends in Cognitive Sciences

469

379

View full text Add to dashboard Cite

Section: Generic Consciousness Combined With Unconscious Iconic Memorymentioning

confidence: 99%

Perceptual consciousness overflows cognitive access

Block

2011

Trends in Cognitive Sciences

469

379

View full text Add to dashboard Cite

“…It was argued that these estimates (e.g., eight objects for the centroid, and seven for faces) exceeded the expected limits of object-based parallel attention (Oksama & Hyönä, 2004;Pylyshyn & Storm, 1988;Scholl, 2001), suggesting that subjects did not rely on this type of object-based sampling strategy. Because previous methods have relied on the observed discrimination thresholds or response variability when subjects must represent, for instance, the size of a single item as a way of estimating the internal noise that might affect ensemble processing (Alvarez & Oliva, 2008;Haberman & Whitney, 2009, 2010, the question of whether the internal noise that affects ensemble representation is lower than the internal noise that affects individual-object representation has not yet been tested. The opportunity has arisen to use variance summation modeling to estimate both the internal noise and the sample size that affect ensemble feature processing from within the ensemble feature task itself.…”

mentioning

confidence: 99%

“…This summary representation or statistic has been called an "ensemble feature." For example, laboratory experiments suggest that humans can represent with little effort the mean size (Ariely, 2001;Chong & Treisman, 2003;Im & Chong, 2009), general direction of motion (Watamaniuk & Duchon, 1992), general location or "centroid" (Alvarez & Oliva, 2008), and approximate number of items in a group (Halberda, Sires, & Feigenson, 2006), as well as the average emotion of faces in a crowd (Haberman & Whitney, 2009, 2010.…”

mentioning

confidence: 99%

“…This too suggests that ensemble features might be represented without selecting individual items. And, for representing the average emotion of a collection of faces (Haberman & Whitney, 2009, 2010 and the average location of a set of dots (Alvarez & Oliva, 2008), measuring subjects' response variability for recalling individual items has been used to estimate the internal noise that might affect ensemble processing and, via Monte Carlo simulation, to estimate the number of items that would need to be averaged in order to attain human-like levels of performance in ensemble feature tasks. It was argued that these estimates (e.g., eight objects for the centroid, and seven for faces) exceeded the expected limits of object-based parallel attention (Oksama & Hyönä, 2004;Pylyshyn & Storm, 1988;Scholl, 2001), suggesting that subjects did not rely on this type of object-based sampling strategy.…”

mentioning

confidence: 99%

See 1 more Smart Citation

The effects of sampling and internal noise on the representation of ensemble average size

Halberda

2012

Atten Percept Psychophys

109

View full text Add to dashboard Cite

Increasing numbers of studies have explored human observers' ability to rapidly extract statistical descriptions from collections of similar items (e.g., the average size and orientation of a group of tilted Gabor patches). Determining whether these descriptions are generated by mechanisms that are independent from object-based sampling procedures requires that we investigate how internal noise, external noise, and sampling affect subjects' performance. Here we systematically manipulated the external variability of ensembles and used variance summation modeling to estimate both the internal noise and the number of samples that affected the representation of ensemble average size. The results suggest that humans sample many more than one or two items from an array when forming an estimate of the average size, and that the internal noise that affects ensemble processing is lower than the noise that affects the processing of single objects. These results are discussed in light of other recent modeling efforts and suggest that ensemble processing of average size relies on a mechanism that is distinct from segmenting individual items. This ensemble process may be more similar to texture processing.

show abstract

“…In particular, just as one averages stimuli far away from the focus of attention in visual perception, one similarly averages stimuli "far away" (i.e., the future rewards of late decision steps) from the current focus of decision in multistep decision making. In line with this proposal, humans can extract ensemble statistics from a variety of features and stimuli, including size (21,22), speed (23), position (24), spatial orientation, and frequency, even under reduced attention (25). Furthermore, ensemble averaging is not limited to low-level perceptual stimuli.…”

Section: Discussionmentioning

confidence: 90%

Continuous track paths reveal additive evidence integration in multistep decision making

Calderon

Dewulf

Gevers

et al. 2017

Proc. Natl. Acad. Sci. U.S.A.

View full text Add to dashboard Cite

Multistep decision making pervades daily life, but its underlying mechanisms remain obscure. We distinguish four prominent models of multistep decision making, namely serial stage, hierarchical evidence integration, hierarchical leaky competing accumulation (HLCA), and probabilistic evidence integration (PEI). To empirically disentangle these models, we design a two-step reward-based decision paradigm and implement it in a reaching task experiment. In a first step, participants choose between two potential upcoming choices, each associated with two rewards. In a second step, participants choose between the two rewards selected in the first step. Strikingly, as predicted by the HLCA and PEI models, the first-step decision dynamics were initially biased toward the choice representing the highest sum/mean before being redirected toward the choice representing the maximal reward (i.e., initial dip). Only HLCA and PEI predicted this initial dip, suggesting that first-step decision dynamics depend on additive integration of competing second-step choices. Our data suggest that potential future outcomes are progressively unraveled during multistep decision making.multistep decision making | computational modeling | reaching task I magine leaving your house in search of food in the neighborhood. Outside, you must first decide to go left or right. Going left subsequently affords a second left-right choice between Thai and Italian food, whereas going right affords another left-right choice between Mexican and Lebanese food. This illustrates a typical twostep tree path decision-making scenario (i.e., four potential tree paths; see Fig. 1A). Such two-step decisions have been conceptualized within the framework of model-based reinforcement learning (1, 2), and recent work has focused on which brain areas underpin reward representation in multistep decision making (3-5). However, the computations underlying multistep decision making are still debated. To address this issue, we distinguish four computational models. We derive and contrast empirical predictions from the four models and test them. In the following paragraph we explain the common ideas and distinguishing features of the four models.In each model, each tree path is associated with an evidence (E) accumulator (e.g., in Fig. 1A, there are four tree paths; we will use Fig. 1A and Supporting Information, Appendix A: Computational Models, Fig. S1, to illustrate the four models). The two leftmost E accumulators (i.e., leading to 3 and 9 in Fig. 1A) are taken as inputs to a left motor evidence (ME) accumulator. The two rightmost E accumulators project to a right ME accumulator. All models reach decisions by gradually updating their E and/or ME accumulator values at each iteration depending on the rewards associated with each tree path. Models can be conceptually distinguished based on three features. The first featuremapping-defines how E accumulators map to ME accumulators; in particular, this feature distinguishes models where only the maximally active E accumulator pro...

show abstract

The Representation of Simple Ensemble Visual Features Outside the Focus of Attention

Cited by 350 publications

References 21 publications

Perceptual consciousness overflows cognitive access

Perceptual consciousness overflows cognitive access

The effects of sampling and internal noise on the representation of ensemble average size

Continuous track paths reveal additive evidence integration in multistep decision making

Contact Info

Product

Resources

About