Dominic Gregory works at the University of Sheffield, in the UK. He has published research—including a book, Showing, Sensing, and Seeming, and numerous papers—on mental imagery, pictures, and other forms of ‘imagistic’ representation, seeking to relate many of their distinctive characteristics to the special nature of the contents that they possess. He has also worked on the logic, metaphysics, and epistemology of modality, on issues concerning perceptual content, and on questions relating to fictional language.
This week at The Junkyard, we’re hosting a symposium on Bence Nanay’s recent book Mental Imagery (Oxford University Press, 2023). See here for an introduction from Bence. Commentaries and replies will follow Tuesday through Thursday.
* * *
Commentary from Dominic Gregory
The current book develops a powerful and wide-ranging case for the importance of mental imagery throughout the mental realm, one that builds upon a distinctively non-phenomenological—and instead neurofunctional—conception of what mental imagery is most fruitfully understood to be. It is written in Bence’s characteristically straightforward but stylish prose, and it is packed with interesting arguments that are richly informed by relevant empirical work, arguments whose conclusions are brought to bear upon a wide variety of concerns.
As remarked, the book’s project is guided by a certain conception of mental imagery: ‘perceptual representation that is used in perceptual processing that is not directly triggered by sensory input’ (p.6). The resulting work falls in the tradition of research by philosophers, psychologists, neuroscientists, and others that aims primarily better to understand how real minds really work; Bence’s distinctive contribution to those efforts is then to build, in the light of a sophisticated blend of characteristically philosophical and detailed empirical observations, a powerful case for the central role within the workings of the human mind of perceptual representations that are not directly triggered by sensory inputs.
Other philosophical work on mental imagery has been guided by alternative aims. That which I know best—my own work on the topic—has focused on using mental imagery as a case-study in theorising about the nature of content; as providing a domain whose study can shed light upon the supposedly special nature of a realm of potential meanings that can belong to appropriate forms of representation. The resulting ideas will hopefully be useful in the study of the actual workings of mental phenomena, of course, but the paths leading to explanations of actual empirical phenomena will generally be somewhat indirect. Other philosophers have focused on the lessons that mental imagery can teach us about mental phenomenology. The differing emphases of the resulting enterprises tend to correspond to conceptions of ‘mental imagery’ that appear to diverge in some ways—perhaps at their cost!—from that explored in Bence’s book.
While questions about the nature of content are not the book’s primary concern, mental images are representational on pretty much everyone’s understanding, a fact that explicitly shapes Bence’s conception of them. And, as the nature of the contents belonging to mental images will surely bear upon their roles in psychological explanations, it is unsurprising that the book does contain some very interesting commentary on the nature of the contents that mental images possess. The rest of this post will focus on what it says about all that.
Chapter 6 of Mental Imagery contains some of the book’s most extensive discussion of the nature of mental imagistic contents. The view that it develops falls quite naturally out of the approach to mental imagery that drives the text. While mental images are taken to be perceptual representations that are not directly produced by sensory input, the absences that provide them with their distinctive genetic character are upstream of the occurrences of ‘subpersonal early cortical perceptual representations’ (p.47) that provide imagery—and perception proper—with its distinctively perceptual nature. The truly perceptual character of mental imagery thus rests on the same basis as the truly perceptual nature of perception itself: the involvement of appropriate subpersonal early cortical representations.
We see, then, that
the content of overall perceptual states depends on the content of these subpersonal early cortical perceptual representations. Similarly, the content of mental imagery also depends on the content of these early cortical perceptual representations (p.46).
Given this shared relationship to a common basis, it is tempting to conclude that the contents of perceptual states and the contents of mental images are basically of the same kind, although maybe sometimes differing with regards to features like their relative determinacy. And, in a later chapter, Bence accordingly writes that
[p]erceptual states attribute properties to the perceived entity. … Mental imagery attributes properties to the imagined entity. While the entities these properties are attributed to are very different (one is imagined, the other is not), the properties attributed to them (and, crucially, the way they are attributed) are similar (p.72).
The previous quotation sketches an attractively simple view of mental imagistic content, one on which the shared neural bases of real perception and mental imagery lead them to share a single sort of content. But other researchers have developed accounts of mental imagistic content that put it at a further remove from perceptual content proper. One (to me, bafflingly popular!) treatment regards visualisations, say, as always representing the visualised scenes as experienced by an implicit subject; Martin and Peacocke are well known for pushing this line. Another—my own view, elaborated in Gregory (2013)—treats mental images as showing things as looking, or sounding or … certain ways, where that comes in two modes: on one, the images characterise sensory episodes from the inside; on the other, they simply characterise worldly scenes. The resulting contents then directly echo perceptual ones in some special respects, but they do not just repeat them.
Do these differences merely reflect differing underlying conceptions of mental imagery? Or should Bence’s simpler treatment of mental imagistic content rule the day—particularly given that all the real-world examples of mental imagery invoked by Martin, Peacocke, me etc. seem likely to feature mental imagery in Bence’s sense?
Here it is worth noting some gaps in the earlier line of reasoning that sketched a path from Bence’s distinctive conception of mental imagery to a broader account of mental imagistic content. One striking feature of Bence’s conception is how it groups together a range of episodes and states that many other researchers have tended to hold apart. It covers standard examples of conscious voluntary imaginative visualisation, for instance, along with the visual images that we summon in memories. But it also encompasses hallucinations—states that may be, from the inside, indiscernible from perceptions proper—and a wide range of non-conscious states as well. (See chapter 5 of Mental Imagery for a nice overview of the terrain.)
Bence argues forcefully that the breadth of his conception of mental imagery is a virtue, making for greater explanatory unification (p.33). Yet that breadth dramatises an important point: that the relationship that mental images and perceptions share to relevant early cortical phenomena is compatible with an extraordinary diversity of phenomenological features. Similarly, though, it is unclear why the common relationship should be taken to ensure that mental images and perceptions possess a single kind of content. Why should the shared relationship be thought to ensure any more uniformity of content than of phenomenology? Maybe hallucinations and perceptions have contents of the same sort, for instance, while the mental images figuring in conscious voluntary imaginings have contents that are somewhat perceptual but that are in other respects essentially different from the contents of real perceptions.
Bence does in fact himself comment that ‘[j]ust how the overall mental imagery representation is put together from the subpersonal representations of early cortical perceptual processing is a complicated question’ (p.47). And that seems to be spot-on: the background conception of mental imagery that Bence is employing does not, it seems to me, dictate any specific account of the contents belonging to the various familiar but narrower kinds of mental images that it covers. For the defining role of early cortical perceptual representations within Bence’s mental images does not itself generate contents for the latter, as there is no requirement for a single story concerning how mental imagistic content overall relates to the contents of whatever early cortical states the relevant mental imagery involves. The book’s background conception of mental imagery thus does not rule out, say, my own view of the contents of (just some of Bence’s) mental images, nor the one endorsed by Martin and Peacocke.
Yet, even if the book’s background conception of mental images does not itself produce an account of mental imagistic content, important questions remain. Why go beyond the sort of minimal treatment that Bence suggests, on which perceptions and mental images have contents of the same broad sort? Is any more complicated approach required? Those are fair queries but not ones that can be tackled here, in a blog post that has already outrun its bounds. To conclude, then, it merely remains for me to thank Bence for having written his insightful and stimulating book.
Response to Gregory
Dominic Gregory’s work, which has greatly influenced me, is very much in the same ballpark as mine. But this, of course, highlights the smaller disagreements. An analogy that comes to mind is from Monty Python’s Life of Brian, where the People’s Front of Judea has many problems with the Romans, but not as much as it has with the Judean People’s Front. So, unsurprisingly, I found Dominic’s comments to be especially helpful.
Dominic’s comments focus on the content of mental imagery: a topic I was trying hard to say as little as possible in the book – partly because others (including Dominic himself, but also Peter Langland-Hassan, among others) have. And partly because I think I can remain noncommittal about the details of an account of the content of mental imagery.
There is one claim about the content of mental imagery, however, that I am very much committed to. It is that the content of mental imagery is of the same kind as the content of perceptual states – after all, mental imagery is a form of perceptual representation (perceptual representation that is not directly triggered by the sensory input). You might wonder: who would deny this? And Dominic certainly wouldn’t.
But some would. Proponents of the so-called ‘Dependency Thesis’ would say that visualizing x amounts to representing the seeing of x. This is not the place to criticize this view, but note that according to this line of thought, the content of mental imagery would be very different from perceptual content (which allows some of the proponents of the dependency thesis to do away with perceptual content while holding onto the content of mental imagery).
Dominic himself has argued very convincingly for a specific version of the view where the content of mental imagery ‘echoes’ perceptual content. This would be fully consistent with everything I say in the book. But – and this is my only disagreement with Dominic’s comments – my view could not be consistent with the Dependency Thesis. So, to return to the Monty Python analogy, the Dependency Thesis is the Romans and Dominic and I are two different People’s Front groups.
Dominic asks whether I am committed to the view that the content of mental imagery and perceptual content are exactly the same (a view he – rightly – rejects). And the answer is that I am not. In fact, I can’t be. Assuming (as I do) some version of intentionalism, the view that perceptual phenomenology supervenes on the content of perceptual states, given that the phenomenology of imagery is different from the phenomenology of perception, the content of mental imagery must also be different from perceptual content. Similar, but not the same. Content of the same kind, but nonetheless, different.
And here it may be helpful to consider the version of intentionalism I argued for recently (in joint work with Manolo Martinez), which we call Many-to-One intentionalism. The idea is that the perceptual phenomenology supervenes on the content of many many perceptual representations (like, for example, representations in the primary visual cortex). So the supervenience relation intentionalism posits is not a one-to-one relation (between perceptual phenomenology and one postulated perceptual content), but a many-to-one relation (between perceptual phenomenology and the content of a large number of perceptual representations). Just how the content of these many perceptual representations (most of which are unconscious) combines is an open question.
And the same is true of mental imagery: just how the content of these many perceptual representations (most of which are unconscious) combines into mental imagery is also an open question. As is the relation between the two.
Thanks to advances in neuroscience and perceptual psychology, we understand more and more about how, for example, the primary visual cortex represents contours. We also understand more and more about how, for example, the primary auditory cortex represent pitch. We also begin to decipher how these representations interact. But a lot more needs to be done to explain how a range of such early cortical representations bring a perceptual state about (and often one with vivid phenomenal character). That could be a book project on its own. And then (and only then) would we be in a position to survey the similarities and differences between the ways these representations combine in the case of perception and in the case of mental imagery.