A Post by Derek Brown.

1. Make-perceive

I am in my office, staring at its off-white walls, and imagining how the walls would look with more photos of my son on them. Impressive? Seemingly not. Briscoe (2008) calls cases like this, where imaginings are placed into the spatiotemporal region of occurrent perceptions, make-perceive. In this example, the imagined photos have some notable features:

• I deliberately constructed them and can make them stop (Deliberate)

• They don’t last long, unless I maintain focus (Fading)

• They are phenomenally present in an elusive, almost invisible way (Phenomenally present as absent, Macpherson 2010)

In addition, it is reasonable to suppose that:

• The imagined photos are caused by my thoughts about those photos (Cognitive source)

2. Make-perceive on steroids?

It has been argued that imagination can impact perception in more robust and systemic ways. For this post I will focus on two phenomena:

• cognitively penetrated colour perceptions or ‘memory colour’ effects (Macpherson 2012)[1]

• amodal completions (Nanay 2010; Briscoe 2011, 2018)

The former is roughly when the experienced colour (e.g. the saturation) of a perceived thing is affected by background beliefs, for example experiencing a Coke bottle to be more red than a similarly coloured red cube, and in Scotland one might experience Irn Bru as more orange than a similarly coloured orange cube.[2] Amodal completion is roughly when partly occluded objects are experienced as ‘completed but partly occluded’ as opposed to ‘a collection of object parts’. There is a pen lying horizontally across my phone.  I experience my phone as ‘a phone with a pen lying across it’ as opposed to ‘two phone parts with a pen between them’.

One can and should scrutinize these phenomena. I will presume they are genuine. I also presume they are perceptual in the sense that they aren’t merely cognitive: they aren’t post-perceptual judgements about what is perceived. Instead, they have a sensory aspect not found in cognition.

Why or how does Irn Bru look more orange than the cube and my phone look to be in one piece? It seems that perception alone can’t explain the effects. Perception can’t explain Irn Bru’s colour because Irn Bru isn’t more orange than the cube. Perception can’t explain why my partly occluded phone looks complete because I’m not getting any visual information from the part of my phone underneath the pen and we can’t see what is hidden from view.

Perhaps imagination can help. Imaginings are typically sensory in a way that cognitions aren’t, so imagination is a candidate to explain the sensory aspect of memory colour effects and amodal completions. On Macpherson’s proposal, Irn Bru is experienced as more orange because you believe that Irn Bru is orange (a cognitive state), and that triggers an imaginative state of orangeness that is injected into the occurrent perception of Irn Bru. Nothing of this sort happens regarding your perception of the cube. As a result, Irn Bru is experienced as being an overly saturated orange. On Nanay’s proposal, I avoid visually experiencing my phone as being split in two by my pen because my imagination projects occluded phone bits behind the pen, and unifies them with the presented (non-occluded) phone bits. Should we posit imagination to explain these phenomena? It’s complicated, and space is short, so let me whet your palate.

3. Two conceptions of perception

At least some basic forms of amodal completion seem to occur in early vision (V1) and occur solely in response to sensory stimulation. Isn’t this therefore just a facet of visual perception, instead of perception that is injected with imagination (Briscoe 2011, 2018)? In response, one might re-assert that we can’t perceive aspects of the world that we aren’t receiving information from. Since this is true of relevant instances of occlusion, we can’t perceive those occluded aspects (Nanay 2010). Here there is a conflict between two competing conceptions of what makes something visual perception:

• Early Processing Conception: early visual processes that are triggered by retinal stimulation must be perceptual.

• Sensory Information Conception: if information about a part of one’s environment isn’t being received by one’s senses, then one can’t perceive that part.

The latter conceives of perceptual systems roughly as informational receivers (and perhaps selectors and transmitters). But many of us, following Helmholtz and even Kant, think of perceptual systems as much more active than this, as making inferences based on systemic assumptions whose outputs go beyond what is received from stimuli. From this perspective, the Sensory Information Conception reflects an oversimplified view of perception. The Early Processing Conception is one way of accommodating more active perception, for it draws boundaries around perception and is mute on what happens within those boundaries. By contrast the Sensory Information Conception draws a line between what is perceivable and what isn’t. Given the plausibility of the Early Processing Conception, suppose it is true. One might think that it tells against the imagination hypothesis for amodal completion. There are at least two reasons to resist this.

First, why not characterize ‘active’ parts of perception in terms of imagination, even if they occur in early stages of vision and are triggered by retinal stimulation? If early vision utilizes stored assumptions and inferential processing on incoming sensory information, yielding outputs that go beyond what is in that incoming information, then something generative and substantive is being contributed by the perceiver, and it is directly impacting the resulting perceptual states and experiences. Perhaps this is a contribution from imagination (Brown 2018).

Second, most forms of amodal completion are not achieved merely in early vision but involve higher levels of activity. Even advocates of the Early Processing Conception have to postulate something extra-perceptual to explain higher-level cases. Since amodal completion is sensory in a way that post-perceptual judgement isn’t, we need something non-cognitive and non-perceptual. Why not imagination (Briscoe 2011, 2018)?

Key parts of this dialectic about amodal completion apply to memory colour effects. The extra orange experienced on/in Irn Bru isn’t coming from Irn Bru. It’s coming from the perceiver. Hence, if one accepts the Sensory Information Conception, something non-perceptual is needed to explain the phenomenon. However, in this case the Early Processing Conception is largely irrelevant to the debate: since memory colour effects involve stored information about complex concepts like Irn Bru, we have good reason to suppose that there are no instances of the phenomenon that can be explained via lower vision.

Tentative conclusions:

(1)   Imagination is a candidate explanation of memory colour and of most instances of amodal completion (i.e. instances of amodal completion that involve contributions outside low-level vision).

(2)   Imagination is a candidate explanation of instances of amodal completion processed in early vision due to the ‘active’ nature of that processing.

4. Dramatic cliff-hanger

To some, (2) is implausible. According to reigning theories of mental imagery in vision science, mental imagery essentially involves higher levels of processing (e.g. Kosslyn et. al 2006). This is in part because that’s the kind of activation we see when we study people who are instructed to (e.g.) ‘Imagine walking into your flat’ and then asked ‘What is in the first room on the left?’ We also think higher-levels of processing are involved because information about one’s flat isn’t stored in lower-vision.

To some, (1) is also implausible. Why should we infer that imagination plays a role in memory colour, or in ‘high-level’ instances of amodal completion? The idea is supposed to be that cognitive systems trigger imaginings, which then get injected into the burgeoning perceptual states. But why not cut-out the middle step and just say that cognition effects the burgeoning perceptual states? The original reason was principally because cognitions aren’t sensory, and so are not of the right form to be the extra orange in experiences of Irn Bru or visually complete the presented phone parts. But cognition doesn’t have to be those extra bits, it just has to cause them, that is, it has to trigger perceptual systems to directly create and be those extra bits (this is how I read Gregory 2017).

On top of this, memory colour effects and amodal completions don’t have the profile of typical cases of imagination, or of typical cases of make-perceive. They aren’t deliberate. They don’t fade. Regarding phenomenal presence as absence, amodal completion seems in the very rough ballpark, but memory colour effects are arguably phenomenally present as present. That is, the experienced orange of Irn Bru is more saturated or richer than the experienced orange of the cube. Still, the match with typical cases of imagination and of make-perceive isn’t great. So perhaps there aren’t good reasons to posit imagination in either case.

I think this is the wrong conclusion (Brown 2018). I don’t think imaginings have to be deliberate or fade or be phenomenally present as absent, in part because there are credible, fairly widespread instances of imaginings that don’t have these features. Like Kant, I have at least sympathy for postulating imagination to explain the active role of perception, given how robustly active I think perceptual systems are. I also don’t think our conception of mental imagery should be overly constrained by experiments where subjects are asked to close their eyes, imagine things, and then answer questions about what they’ve imagined. If we are asking whether perceptions are often infused with imaginings, then we need a conception of imagination that applies when subjects’ eyes are open and they aren’t being asked to imagine things and answer questions about what they’ve imagined.

Where does this leave us? I suspect that many aspects of our perceptual states and experiences are partially imaginative in some sense but might not tick ‘enough boxes’ in the imagination column to warrant being imaginative full stop. As a conclusion this won’t strike many as satisfying. However, it isn’t a conclusion. It’s an orientation from which to work through more of the details. I don’t think we have catalogued the properties of imagination and the perceptual phenomena that might have imaginative aspects and constructed a framework that reveals which phenomena arguably have imaginative aspects, and in which ways. Nor have we formulated a conception of perceptual states and experiences with this in mind.[3]

[1] There is a vast literature on cognitive penetration, including a decent amount of healthy skepticism. I focus on colour memory effects for numerous reasons, including because the data in favour of it is now fairly robust, colour perception is a basic or fundamental aspect of vision, and the phenomenon arguably cannot be explained by other perceptual things like attention. In general, I only include philosophical references. These contain numerous relevant scientific ones. One important criticism of memory colour is Gross et al. (2014).

[2] By ‘similarly coloured’ I mean an object that that is objectively the same colour as the target object (i.e. it is physically identical colour-wise to the target object).

[3] Thank you to Robert Briscoe and Fiona Macpherson for helpful comments on an earlier draft of this post, and apologies to Robert for not having the space to incorporate them all.

References

Briscoe, R. 2008. Vision, action and make-perceive. Mind and Language, 23: 457–497.

--       2011. Mental imagery and the varieties of amodal perception. Pacific Philosophical Quarterly, 92: 153-173.

--       2018. Superimposed mental imagery: On the uses of make-perceive. In F. Macpherson and F. Dorsch, eds., Perceptual Imagination and Perceptual Memory, pp. 161-185. Oxford University Press.

Brown, D. 2018. Infusing perception with imagination. In F. Macpherson and F. Dorsch, eds., Perceptual Imagination and Perceptual Memory, pp. 133-160. Oxford University Press.

Gregory, D. 2017. Vision and Visual Imagery. The Junkyard.

Gross S. Chaisilbrungraung, T. Kaplan, E. & Menende, J. A. 2014. Problems for the purported cognitive penetration of perceptual color experience and Macpherson’s proposed mechanism. The Baltic International Yearbook of Cognition, Logic and Communication, vol. 9, Perception and Concepts: 1-30.

Kosslyn, S., Thompson, W. and Ganis, G. (2006). The Case for Mental Imagery. Oxford, UK: Oxford University Press.

Macpherson, F. 2010. Phenomenal presence as absence. Lecture given at a conference on Phenomenal Presence in Fribourg, Switzerland, June 2010.

--       2012. Cognitive Penetration of Colour Experience: Rethinking the Issue in Light of an Indirect Mechanism. Philosophy and Phenomenological Research 84: 24-62.

Nanay, B. 2010. Perception and imagination: Amodal perception as mental imagery. Philosophical Studies, 150:  239-254.