Fitting perception in and to cognition

1 Fitting perception in and to cognitionRobert L. Goldstone , Joshua R. de Leeuw, David H. LandyIndiana University, Department of Psychological and Brain Sciences, Program in Cognitive Science, 1101 E. 10th Street, Bloomington, IN 47405, United Statesarticle infoArticle history:Available online 8 December 2014 Keywords:Perceptual learningCategorical perceptionModularityCognitive penetrabilityCognitionPerceptionabstract Perceptual modules adapt at evolutionary, lifelong, and moment-to-moment temporalscales to better serve the informational needs of cognizers. Perceptual learning is a power-ful way for an individual to become tuned to frequently recurring patterns in its specificlocal environment that are pertinent to its goals without requiring costly executive controlresources to be deployed.

2 Mechanisms like predictive coding, categorical perception , andaction-informed vision allow our perceptual systems to interface well with cognition bygenerating perceptual outputs that are systematically guided by how they will be classic conceptions of perceptual modules, people have access to the modules outputsbut no ability to adjust their internal workings. However, humans routinely and strategi-cally alter their perceptual systems via training regimes that have predictable and specificoutcomes. In fact, employing a combination of strategic and automatic devices for adaptingperception is one of the most promising approaches to improving cognition .

3 2014 Elsevier All rights IntroductionAttempts to describe the interface between perceptionand cognition presume that perception and cognition canbe separated from one another. It only makes sense to talkabout the interface between processes A and B if they are,in fact, two separate, albeit linked, processes. Given the dif-ficulties in sharply delineating between perception andcognition, some thinkers have been led to the radical moveof completely lumping them together. With the notion ofperception as unconscious inference,Helmholtz (1867)joined perception and cognition , with both cruciallyinvolving the interpretation of the world. The Buddhistmental factor Sam j a has been translated alternatively as cognition or perception .

4 Talmy (2000)advocated usingthe term ceptions to purposefully merge perception andconception, motivated by an effort to break down artificialboundaries between these mental acts. More recently,Clark (2013)has argued that the lines between perceptionand cognition [are] fuzzy, perhaps even vanishing (p. 190).2. Adaptive perceptual modulesStill, there are good reasons to prefer construing per-ception and cognition as interactive, even overlapping,processes, but nonetheless differentiated. The brain iscomprised of anatomically localized regions with relativelydense within-region neural connectivity and sparserbetween-region connectivity.

5 Brain regions can often timesbe attributed specific perceptual tasks, such as the percep-tion of color, binocular depth perception , reading, and facerecognition. The articulation of the brain into modulessuch as these is crucial for achieving fast and reliable per-ception (Nakayama, 2005). Cases of cognitive impenetra-bility exist in which particular goals, expectations, andbeliefs one has do not influence one s perception (Pylyshyn, 2003). More generally, there is a theoreticaladvantage to conceptualizing most active agents embed-ded in environments in terms of perception , cognition ,and action. For extended and embedded agents from 2014 Elsevier All rights reserved.

6 Corresponding author. Tel.: +1 812 855 Goldstone). cognition 135 (2015) 24 29 Contents lists available atScienceDirectCognitionjournal homepage: people to webpages to aircraft carriers, it is useful tothink of some components playing primarily informationprocessing roles, and some as primarily processing the spe-cific nature of the environmental information. Our concep-tualization of conceptualization itself should not becomeso blended with perception that the evidence for andconceptual advantages of partially independent perceptualmodules are the existence of perceptual modules does notcommit one to the assumption that these modules arehardwired or fixed in their function.

7 In fact, perceptualmodules are highly adaptive and become attuned acrossseveral temporal scales to the cognitive needs of an organ-ism. At the longest, evolutionary time scale, organismsevolve perceptual systems that are tailored to their stableenvironment. A compelling case of this is the close matchbetween the peak light wavelength sensitivity of photore-ceptors in fish to the most prominent wavelengths in theirwater environments (Lythgoe, 1972). At the intermediatetime scale of learning throughout an organism s lifetime,perceptual modules become tuned to frequently recurringpatterns. At the most rapid time scale of moment-to-moment changes in context, responses in our earliest per-ceptual systems become modified by expectancies (Lupyan& Ward, 2013).

8 For example, training in a selective atten-tion task produces differential responses as early as thecochlea (Puel, Bonfils, & Pujol, 1988). This amazing degreeof top-down modulation of a peripheral neural system ismediated by descending pathways of neurons that projectfrom the auditory cortex all the way back to olivocochlearneurons, which directly project to outer hair cells withinthe cochlea an impressively peripheral locus learning over the course of an organism slifetime is a particularly powerful way of altering the func-tioning of perceptual modules so that they come to betterserve an organism s cognitive needs. Even if humans arenot consciously and strategically changing the wiring ofperceptual modules (a possibility we will return to later),these modules nonetheless adapt systematically at thetime scales of tens to thousands of repetitions to allowan organism to better make discriminations and categori-zations that are vital to its interests.

9 Empirical evidencepoints to neurophysiological changes to properly percep-tual, rather than post-perceptual decision, brain example, when monkeys are trained with one of twovisual discrimination tasks (a bisection task or vernier dis-crimination), their primary visual cortex (V1) neurons takeon different novel function properties pertinent to thesetasks even when presented with the same shape (Wu,Pi ch, & Gilbert, 2004). These V1 differences are observablefrom the very earliest neural responses following stimulusonsets. Generalizing over many studies, training in bothauditory and visual tasks produces early changes to manyperceptual modules.

10 One of the mechanisms for thesechanges are that neurons become more selective in theirresponses and the cortical representations of different fea-tures become increasingly less overlapping (Crist, Li, &Gilbert, 2001).This evidence from neuroplasticity apparently clasheswith epistemological concerns about perceptual systemsbeing tainted by preconceptions. The concern is that ifour perception of the world depends on our experiencesand wishes, then how can these perceptions then provideus with unbiased evidence about the world (Siegel,2012)? AsFodor (1983)puts it: seeing what we expectseems to defeat the purpose of vision: [an organism] gen-erally sees what s there, not what it wants or expects tobe there.