83

What we need to think about when wethink about unconscious perception Ian Phillips St. Anne’s College, University of Oxford, Oxford OX26HS, UK. E-mail: ian.phillips@st-annes.ox.ac.uk Theoretical discussions of unconscious perception typically focus on how consciousness should be operationally defined (Lau 2008; Seth et al. 2008; Irvine 2013). However, a compelling case of unconscious perception requires both evidence that consciousness is absent and that perception is present. Consequently, theorists must also consider how perception should be operationally defined, and assess alleged cases of unconscious perception accordingly. Traditionally, it was assumed that to be perceived a stimulus must contribute to a subject’s conscious perspective (Moore 1925). To allow for the possibility of unconscious perception, Kanwisher suggests instead using “perception” to refer to “the extraction and/or representation of perceptual information from astimulus, without any assumption that such information is necessarily experienced consciously” (2001, 90). Kanwisher’s proposal needs refinement. It risks counting distinctive allergic reactions as instances of perception (Dretske 2006). It also fails to secure the idea that perception is an individual-level phenomenon, not merely an occurrence in an individual’s visual system or brain (Burge 2010, 368 ff.). By wayofrefinement, Burge proposes that perception is constitutively a matter of objective sensory representation by the individual. This means that perceptual states do not merely carry information but represent features of the physical environment as opposed to “idiosyncratic, proximal or subjective features of the individual” (2010, 397). According to Burge, such contents are attributable just when perceptual constancies are exercised. “Perception requires perceptual constancies.” (399). An alternative approach focuses on the “role”, as opposed to “content”, of perceptual states. Thus, Dretske (2006) proposes that the information which perceptual states carry must be directly available for the control and guidance of action. Similarly, Prinz (2015) stipulates that perception involves the transduction of “useable” sensory information. Content and role approaches are not exclusive. Milner and Goodale understand perception to “refer to a process which [subserves] … the recognition and Downloaded from https://academic.oup.com/nc/article-abstract/2017/1/nix015/4107416 by New York University user on 28 November 2017 identification of [external] objects and events and their spatial and temporal relations” (1995/2006, 2). Here both content and role requirements are in play. In line with contemporary orthodoxy, all the authors just mentioned claim that perception howsoever defined occurs unconsciously. [See also Block (2016) and Block in Block and Phillips (2016).] Here, I discuss four cases commonly invoked in support of this contention. Thinking about whether perception is genuinely present in these cases demonstrates that matters are muchless clear cut than standardly supposed. Case 1: Blindsight According to Burge, “blindsight patients perceive environmental conditions. The perception involves perceptual constanciesincluding motion, location, and size constancies. The perception guides action. There is strong reason to believe that some of these patients lack phenomenal consciousness in the relevant perceptions.” (2010, 374) In short: blindsight constitutes genuine perception without consciousness. It is important to consider the possibility that blindsight in fact involves abnormal and degraded, but nonetheless conscious, vision (Phillips 2016). However, my current interest is in the importance of asking whether residual function constitutes genuine perception. For Burge, a positive answer requires the preservation of visual constancies. Yet constancy preservation in blindsight is far from uncontroversial. Patient DB perceives neither surface color nor chromatic contrast, matching colored stimuli purely on the basis of wavelength (Kentridge et al. 2007; Alexander and Cowey 2013). Motion detection in GY is limited to “objectless” firstorder motion energy (i.e., spatiotemporal changes in luminance) as opposed to changes in position or shape (Azzopardi and Hock 2011). And MS and GY’s capacities to locate and detect objects are arguably limited to the detection of sharp luminance contours and stimulus transients, “‘events’ varying ‘in subjective salience’” not objective environmental features (Alexander and Cowey 2010, 532). Assuming with Burge that constancies are necessary for perception, such findings suggest that the relevant preserved capacities of these patients at least do not constitute perception. (See also Case 4 regarding “actionblindsight”.) Case 2: Subliminal Priming Using a novel metacontrast masking paradigm (Fig. 1), Norman et al. (2014) demonstrate preferential facilitation of color identification responses to a target preceded by an undetectable prime matched in surface color when compared with one matched in reflected color (i.e., wavelength). This suggests that color constancy can occur outside consciousness. Does it evince unconscious “perception”? That turns on how the prime facilitates responding. One possibility is that the prime elicits activity in the visual system, including surface color representations, such that the visual system processes subsequent surface-congruent stimuli more fluently. It is consistent with this understanding that information concerning the prime’s surface color is confined to the visual system and so wholly unavailable for action control and guidance. Consider a crude analogy: we do not perceive everything that causes pupillary dilation, yet such dilation makes various subsequent stimuli easier to perceive. Another possibility is that the prime automatically activates associated responses in the motor system (Schmidt et al. 2006). However, such response priming also does not demonstrate6 | Petersetal. Figure 1. Trial sequence from Norman et al. (2014) Notes: From left to right: a spatial cue and a brief delay precede the presentation of a prime for 0 (prime absent), 12.5, or 37.5ms; a variable delay follows, after which a change in the display’s illumination boundary occurs. This alters the illumination of the region in which prime and target (mask) are presented. Finally, a target (mask) is displayed. This is either green or blue and, given the previous change in illumination, respectively, matches the prime in either “reflected” or surface color. The subject then identifies the color of the target as rapidly as possible. Source: Reproduced with permission. CopyrightV C 2014 Elsevier Ltd. All rights reserved. that the relevant information is available for individual-level action control and guidance. Indeed, evidence suggests that comparable unconscious primes are not available for central, executive control of action (Kunde et al. 2003; Ansorge et al. 2011; cf. the “radically uncontrollable” effects in Snodgrass and Shevrin 2006; Cressman et al. 2013). These concerns have wide scope. They threaten the relevance of much work on subliminal priming to the question of unconscious perception, at least if weadopt arobust role-requirement on genuine perception. Case 3: Unconscious Attention There is now compelling evidence that spatial, feature-based and object-based attention can all operate outside of awareness (e.g., Kentridge et al. 1999, 2004; Schmidt and Schmidt 2010; Norman et al. 2015). However, treated as evidence of unconscious perception, such results merit much the same critique as priming studies: it is one thing for a stimulus to elicit (even widespread) activity in the subject’s brain which (perhaps dramatically) affects subsequent visuomotor processing; it is another thing for a representation of that stimulus to be available for voluntary, individual-level control and guidance of action. Burge claims that attention itself “is a type of psychological agency” and “[h]ence …attributable to the individual” (372; see also Block in Phillips and Block 2016). But, in the present context, attention refers to the stimulus-driven modulation of visuomotor processing. Why should we think of the agent themselves as doing this any more than we should think of the agent as dilating their pupils? Case 4: Vision-for-Action Superficially the strongest case for unconscious perception in the robust senses here in focus, looks to work on so-called vision-for-action representations associated with the dorsal Downloaded from https://academic.oup.com/nc/article-abstract/2017/1/nix015/4107416 by New York University user on 28 November 2017 stream, and in particular to studies of patients with visual form agnosia (Milner and Goodale 1995/2006) and so-called “actionblindsight” (Danckert and Rossetti 2005; Whitwell et al. 2011). Such patients fail to reveal evidence of size constancy in their explicit, for example, manual or verbal, reports. However, their accurate scaling of grasp aperture in visually guided grasping provides evidence that (vergence based) size-constancy is preserved (Marotta et al. 1997; Sperandio et al. 2012; see also MonWilliams et al. 2001; Servos 2006). However, it is far from obvious that vision-for-action constitutes genuine perception. [As ever issues about residual awareness also need pressing. See, for instance Whitwell et al. who write of their patient, SJ: “It is important to note, however, [that her] failure to show a target redundancy effect in our experiment does not mean that she is completely incapable of detecting targets in her blind field (using a button press). Had we used a forced-choice variant of this task she may have very well exhibited better-than-chance levels of performance.” (2011, 915).] Indeed, although Milner and Goodale do think perception can occur unconsciously, they insist that “[t]he visual information used by the dorsal stream for programming and on-line control … is not perceptual in nature” (2008, 776; cf. 1995/2006, 2). From the present perspective, a concern is whether the pertinent modulations of behavior (e.g., grasp aperture) witness genuine control and guidance “by the individual”, and so meet relevant role requirements for perception. An alternative picture is suggested by the familiar metaphors of an automatic pilot (Pisella et al. 2000), tele-assisted semi-autonomous robot (Goodale and Humphrey 1998, §9; Goodale and Milner, 2004, 98101; Milner and Goodale 1995/2006, §8.2.3), or heat-seeking missile (Campbell, 2002, 56). Such metaphors suggest that the only representations attributable to the individual will be those associated with target and action-type selection (cf. Clark 2007, 576). Representations exclusively involved in fine-grained motor programming will not be individually attributable. Indeed,The ASSC20 debate | 7 Danckert and Rossetti note how, as with unconscious priming discussed above, the parietal system “often functions automatically, rapidly modifying visually guided hand movements … in contradiction to conscious commands” (2005, 1042, see also Pisella et al. 2000). Conclusion Proponents of unconscious perception face the challenge of providing an adequately justified operational definition of individual-level perception. Assessed in the light of extant proposals, many apparently clear cases of unconscious perception no longer appear so clear cut. Moreover, an obvious concern lurks in wait. One possible operational test for perception (closely associated with Dretske’s role-based proposal above) requires that the information carried by a perceptual state must be exploitable by a subject to make a discriminatory response. Yet this test for perception is equivalent to so-called “objective” measures of consciousness (i.e., above chance discriminative sensitivity) (Green and Swets 1966). As a result, no putative cases of unconscious perception can hope to avoid the familiar concern that they simply involve weak conscious awareness unreported due to a conservative response criterion (Eriksen 1960; Holender 1986; Phillips 2016; Peters, this symposium). Unconscious perception within conscious perception NedBlock Department of Philosophy, New York University, 5 Washington Place, NewYork, NY 10003, USA. E-mail: ned.block@nyu.edu The debate on unconscious perception– including much of my own contribution– has been misdirected (Peters and Lau 2015; Block 2016; Block and Phillips 2016). The debate has focused on the efficacy of methods designed to shave off the conscious part of a perception, leaving unconscious perception. Can it be done? I think the answer is yes, but the issue is not straightforward. Whether or not it can be done, and whether or not the resulting unconscious percept would have to be different from the unconscious part it resulted from, unconscious perception is ubiquitous since many (if not all) conscious perceptions have unconscious perceptions within them. I will start by explaining how we can distinguish conscious from unconscious representations in the visual system. Then I will move to the issue of whether the difference between these conscious and unconscious representations is the same as the difference between personal and sub-personal representations. I amdiscussing this latter issue because of the issue of whether unconscious visual representation is always sub-personal and hence not genuine perception (Block and Phillips 2016). Howcanwedistinguishbetweenconscious and unconscious visual representation? One useful approach is to focus on neural bases. Every conscious visual perception in an animal has a neural basis and within that neural basis there is a neural basis of unconscious visual representation. If the viewer is far enough away from Fig. 2 so that the stripes of the left and middle discs of Fig. 2 subtend one-fiftieth of a degree or less, all three discs will look the same (see Fig. 3). But He and MacLeod (2001) (see also Smallman et al. 1996) showed that some gratings that cannot be resolved consciously nonetheless are visually represented. The lense of the eye blurs very fine gratings but they were able to project similar grids directly to the retina through the side of the eye, bypassing the lense, by using laser interferometry, showing that gratings that are not consciously seen produce after-effects of the same magnitude as gratings that are consciously seen, and revealing that both gratings are indeed represented in the visual system, including in the retina and early vision. An analogous result obtains for color flicker. If two colors alternate at frequencies above 10Hz (10 cycles per second), “heterochromatic flicker fusion” occurs: viewers consciously see a single fused color rather than flickering colors (so long as the two colors have the same luminance). For example, red and green flickering above 10Hz looks non-flickering and yellow. (Combining red and green lights– an “additive” mixture, as is used in your computer screen– produces yellow.) However, retinal cells respond to flicker way above the frequency that the subject can consciously see– as high as 40Hz, and a way station between the retina and the cortex (the lateral geniculate nucleus) responds to frequencies that are almost as high. In the first cortical visual area, V1 (and probably to a lesser degree V2), all cells responded to 15Hz flicker and most to 30Hz flicker (Gur and Snodderly 1997). In sum, the retina and early vision registers flicker that the subject does not consciously see as flicker. How do we know these representations in early vision are unconscious? Subjects show no sign of awareness of them and are at chance in guessing. For example, Haynes and Rees (2005) were able to predict the orientations that subjects were seeing from activity in V1 at a greater than chance level even when subjects were at chance in choosing which orientation they saw. But are these unconscious flicker and orientation contents really contents of perception or are they just sub-personal representations of information on a par with representations in the brain of autonomic nervous system properties like heart rate or representations in neurons in the gastrointestinal tract? First, the unconscious representations have many of the same Figure 2. These circles look different to you but if you move further away from the screen or page to the extent that the distance from the beginning of one black bar to the beginning of another in the grid subtends one-fiftieth or less of a degree of visual angle they will look the same. Downloaded from https://academic.oup.com/nc/article-abstract/2017/1/nix015/4107416 by New York University user on 28 November 20178 | Petersetal. Figure 3. The point at which the bars merge is about 50 cycles per degree of visual angle, which will differ depending on your distance from this figure. Note: Thanks to Marisa Carrasco for this image. contents as conscious representations. For example, both conscious and unconscious representations can have contents of flicker under 10Hz and orientation of grids under 50cycles/degree. Further, these representations are alike in many respectsfor example with regard to perceptual aftereffects and perceptual constancies. Constancies are important to what counts as perception (Burge 2010). There is strong evidence that size constancy is registered in V1 independently of feedback from higher areas and good evidence that lightness and color constancy is registered in V1 and V2 also independently of feedback (MacEvoy and Paradiso 2001; Hurlburt 2003; Pooresmaeli et al. 2013). While there are clear cases of sub-personal representations (such as gastrointestinal representations) and personal representations (e.g., conscious perceptions), many if not most cases of interest are indeterminate and there is no accepted characterization of the difference. Every proposal that has been made for what the personal/sub-personal distinction comes to has an air of postulation (Block and Phillips 2016). Extreme views abound. For example, according to a widely cited paper by Charles Travis (2004), all perceptual representations are subpersonal. People have intuitions about some cases but those intuitions may concern reasonable postulations about how to use a technical term. By contrast, there is a robust border between conscious and unconscious representations in some domains. As just explained, 50cycles/degree is the border between consciously and unconsciously seen gratings: representations in the visual systemof gratings that are more fine-grained than 50 cycles per degree are unconscious. And alternating representations of color above 10Hz are unconscious. (No doubt there are borderline cases and somevariation from person to person.) In short there is a fairly determinate border between conscious and unconscious, at least in some domains, but no one has given a reason to believe in a determinate border between personal and sub-personal, so there is no case that the two distinctions coincide. Anyone who claims that they do coincide owes us a persuasive way of drawing a boundary between the personal and sub-personal that does not beg the question against unconscious perception. Downloaded from https://academic.oup.com/nc/article-abstract/2017/1/nix015/4107416 by New York University user on 28 November 2017 I have tried to give sufficient conditions for the personal level, conditions that I now regard more as useful postulations than as objective facts about the distinction. In previous publications touching on the personal/sub-personal issue (Block 2016; Block and Phillips 2016), I have focused on three supposed indicators of the personal level: whether the perceptual representations guide the person’s action, whether they engage the person’s preferences or needs, and whether they engage the person’s understanding. I have no space to discuss all of them but I will sketch how the first of them does classify some unconscious perceptual representations as personal– and hence as unconscious perception. The cases I have in mind involve the dorsal visual system (Fig. 4). Mel Goodale and David Milner have extensively tested a brain damaged patient known as DF (or sometimes Dee) who had damage to an area in the ventral visual cortex that underlies form perception (2008). DF could consciously see colors and textures but not shapes or orientations. If shown a slot as in Fig. 5, she was consciously aware of a blob with no orientation and she was nearly at chance in saying what the orientation of the slot was and in matching a card to the direction of the slot as shown onthe left. However, and this is the indication of unconscious perception of orientation, she could nonetheless post the card through the slot with accuracy only slightly less than that of normal subjects. The orientation was represented in her dorsal visual system but her conscious visual system represented colors and textures. (The accuracy depictions in Fig. 5 are “normalized”. That means that all slots are treated as if they were vertical, and when the subject got the estimations and posting wrong, that is graphed as a deviation from the vertical.) How do we know that these perceptions are genuinely unconscious? Why else would DF be nearly at chance on matching and saying what the orientation is? Importantly, this paradigm does not require completely unconscious perceptual states. There is no need to shave off the conscious part of a perception. In this paradigm, there are unconscious perceptual “contents” in otherwise conscious perceptions. DF consciously sees the stimulus but not the orientations of the slot. In the case of DF there is as definitive an answer to the personal/sub-personal question as one is going to get for this question: her unconscious visual representations are “her” representations because they guide her actions in posting. Volker Franz and Thomas Schenk have criticized some of Goodale’s and Milner’s studies (Hesse et al. 2011). They argue that DF may not visually represent orientations at all but rather manage to post by a trial and error procedure, adjusting her posting to avoid hitting the edge of the slot. Their evidence derives from a study in which they gave normal subjects the task of posting a card in a “slot” that was really a rectangle that was much wider than the slot at the top of Fig. 5. The length of the rectangle was shorter than the card so subjects could only fit the card through by putting it on the diagonal. Their subjects did do it in part by avoiding the edges. However, their slot was designed to elicit obstacle avoidance and so does not reflect on behavior involving a slot that is not so designed. Further as Goodale and Milner note, “Dee moved her hand forward unhesitatingly, and almost always inserted the card smoothly into the slot. Moreover, video recordings revealed that she began to rotate the card toward the correct orientation well in advance of arriving at the slot.” (2005, 20) (I have seen some of this footage and would concur for the footage I have seen.) Further, it would seem that orientation perception would be required to avoid hitting the edge of the slot with the card.The ASSC20 debate | 9 Figure 4. The back of the head is on the right; the conscious ventral visual system is represented by the bottom arrow from gray to purple whereas the largely unconscious dorsal visual system is represented by the top arrowfrom gray to green. Figure 5. See text for explanation Note: Thanks to Mel Goodale for this image. Another experiment that combines conscious perception with partially unconscious contents tested normal subjects on gripping “Efron blocks”, blocks that differ in width though with surface area and brightness held constant over the various widths (Goodale and Murphy 1997). They affixed measuring devices to the thumb andforefinger in order to measure grip precision at various angles from fixation of five of these blocks. Subjects also had to say which of the five blocks they were seeing. If you hold out an object at a 70 angle from your line of sight, you will see that you are very dimly conscious of its size: you consciously can tell the difference only between a very large and very small thing. What Goodale and Murphy found was that grip precision did not differ much between 5 and 70 , but numerical judgments of the widths (the blocks were given numbers) had half the precision at 70 as at 5 . This is not surprising given that the conscious ventral visual system that dominates foveal vision is much weaker than the action-guiding dorsal visual system in peripheral vision. The fact that the subjects’ grip was muchmore precise in the periphery of vision than the subjects’ size estimates strongly suggests that the guidance of grip is not entirely conscious. In other words, unconscious representation of widths is partly guiding gripping. That unconscious representation of width is– arguably– at the personal level because it guides action. Note that this argument for unconscious perception requires no shaving off of the conscious part of Downloaded from https://academic.oup.com/nc/article-abstract/2017/1/nix015/4107416 by New York University user on 28 November 2017 vision. What is unconscious is an aspect of the content of perception. To conclude: the debate about the existence of unconscious perception has focused on whether the conscious part of a conscious perception can be shaved off, leaving only an unconscious perception. Whatever the resolution of that issue is, much if not all of conscious perception involves unconscious perception.