Margot Strohminger is Senior Lecturer in Philosophy at Australian Catholic University, Melbourne, Australia. Before that she held a string of postdocs in Europe including a Marie Curie Fellowship and JRF at Oxford. And before that she studied philosophy at McGill (BA), St Andrews (PhD) and Sheffield (MA) inbetween, where she was lucky enough to work with the author of The Profile of Imagining.
This week at The Junkyard, we’re hosting a symposium on Rob Hopkins’ recent book: The Profile of Imagining (Oxford University Press, 2024). On Monday, we began with an introduction from Rob. Commentaries and replies are following Tuesday through today.
* * *
Commentary from Margot Strohminger
How Much of an Alternative to Simulationism?
On a popular philosophical conception, the imagination serves to ‘simulate’ or ‘recreate’ many ordinary mental states and processes. The scope of ordinary mental states and processes that can be simulated is thought to include at least the experiences at work in visual and other sensory perception.[1] It is thus natural to ask how much of an alternative to simulation Hopkins’ account of the sensory imagination represents—a question Hopkins himself briefly addresses in the final chapter of The Profile of Imagining (PI). There Hopkins emphasizes the differences between his profiling account of sensory imagination and a simulationist approach.
In this commentary I will explore the possibility that the profiling account is compatible with, and even a welcome supplementation to, simulationism. We can understand simulationism as providing a schematic theory of the imagination and the profiling account as filling in gaps left open by simulationism for the special case of sensory imagining. Thus the two approaches are not in genuine competition.[2]
According to the simulationist, sensory imaginings resemble sensory experiences in important respects. At the same time they are different enough from the latter that we should think of them as a different kind of mental state altogether (they are imaginings or simulations). Detractors of simulationism may complain that the theory is uselessly vague about what these similarities and differences consist in. In response a simulationist may piggyback on Hopkins’ work. This is because it provides a detailed reckoning of these similarities and differences in the case of sensory imagination.
Consider the phenomenological differences between sensory imagination and sensory experience proper. Simulationists going back to Hume (1975 [1739]: Part 1, §§1-3) emphasize the difference in ‘force’ or ‘vivacity’ among the two: the experience of seeing a crocodile is said to have a ‘forcefulness’ or ‘vivacity’, which its imaginative counterpart lacks.[3] Meanwhile Hopkins develops an entirely different set of phenomenological contrasts. The primary contrasts are as follows (PI, pp. 4-5):
ACCURATE/INACCURATE. Perceptions present themselves as being a guide to how things are, whereas imaginings present themselves as not being a guide to how things are.
RECEPTIVE/SPONTANEOUS. Perceptions present themselves as being determined by things outside of ourselves, whereas imaginings present themselves as arising from within us.
INDEPENDENT/DEPENDENT. Perceptions present their objects as existing independently of ourselves, whereas imaginings present their objects’ existence as dependent on ourselves.
OVERFLOWING/EXHAUSTED. The objects of perception appear to have features of which we are not aware, whereas the objects of imagination appear to not have features outside of our awareness.
I see no reason why a simulationist cannot accept the existence of all of these differences, whether in addition to the Humean contrast in vivacity or not.
Where else might we look for conflict between the profiling account and the simulationist treatment of imagination then? Arguably they offer different explanations for whatever phenomenological resemblance there is between sensory perceptions and imaginings.
Central to the profiling account is, well, profiling: both perceptions and imaginings involve the use of ‘profiling’. Hopkins devotes two chapters to explaining how, which resist quick summary (Chapters 3 and 4). I will content myself with illustrating the key idea. Hopkins writes:
in perception knowledge of sensory profiles, operating on perspectival properties with which we are presented, makes other aspects of the world perceptually present to us. […] the same knowledge, put to work in loosely parallel ways on rather different materials, lies at the heart of sensory imagining. (PI, p. 94)
So, for example, whether I see or merely visualize the scene of my desk, I deploy knowledge of sensory profiles, which are a set of possible perspectival properties that are unique to the non-perspectival properties of the scene (the surface colors and real shapes of its components, e.g.). I deploy this knowledge in two very different ways, however, depending on whether I am having a visual perception or a visual imagining of the scene.
What might the simulationist say about this as a proposed explanation of the resemblance between the two kinds of states? Is it in tension with their own? The core simulationist claim—that mental simulations resemble the states and processes they simulate—does not involve an explanation of what explains or underlies the resemblance at all. So a simulationist need not even be seen as offering such an explanation. Some simulationists though can be seen as offering one: namely, because simulations “re-use” the same cognitive mechanisms as the states and processes being simulated.[4] Meanwhile PI provides the basis for a different explanation: to the extent that the former resemble the latter, this is because of the parallel usage they make of profiling (cf. PI, p. 254). Arguably the two explanations just aired can both be endorsed simultaneously. To do so, take the case of visual imagining as an example. A profiling-friendly simulationist might claim that a subject will deploy knowledge of visual profiles in one of the two ways elaborated by Hopkins, depending on whether they are using or reusing the mechanisms proprietary to visual processing.
I said before that Hopkins’ account treats sensory imaginings. That isn’t quite right. He also sketches how to extend that account to another class of experiential imaginings, ‘affective imaginings’ (Chapter 8). In affective imagining, we imagine what it is like to experience some emotion or other affect. Regardless, one advantage of simulationism might seem to be its generality: it applies to any state or process which we pre-theoretically think of as an imagining (and arguably more besides[5]). This includes much more apart even from experiential imagining: consider propositional imagining or imagining-that. I am not convinced this generality is a genuine advantage for the simulationist over the profiling theorist as, again, we might just see profiling as relevant to some simulations but not others (with resemblance and/or re-use being relevant to all of them).
Hopkins’ discussion of these differences also might help simulationists get clearer on what class of sensory experiences their object of study are simulations of: just veridical sensory experiences? What about hallucinations? Mental imagery present in dreams? Hopkins reserves the term “clear-sighted imaginings” for those sensory imaginings which have the four phenomenological features above: INACCURATE, SPONTANEOUS, DEPENDENT, and EXHAUSTED. He is open to counting hallucinations and dreams as sensory imaginings as well and he sketches some options for how to define sensory imagining so that they will or won’t end up counting (PI, pp. 145-7). Suppose we take the accessibility of the latter phenomenology to be essential to sensory imaginings and further that it turns out that this phenomenology is in principle accessible to the hallucinating or dreaming subject. Then hallucinations and dreams will count as sensory imaginings as well. Or perhaps their phenomenology is closer to the phenomenology of veridical sensory experiences. Then they will not. The simulationist also can arguably go either way here as well, although admittedly it is unclear whether their treatment of these borderline cases will happen to align with the options sketched by Hopkins. In any case The Profile of Imagining should spur the simulationist to fill in crucial details in their framework and those details could even end up borrowing heavily from its rich insights.
Notes
[1] For overview see e.g. Barlassina and Gordon 2017.
[2] Cf. Hopkins: “If the recreativist is content with this functional parallel [namely, the role of profiling in both sensory perceiving and imagining], who am I to argue? If recreativism is to have any content, some such parallels must be spelled out; and if the views developed in this book help in that task, so much the better” (PI, 254).
[3] See Kind 2017 and Langkau 2021 for recent discussion of how to understand vivacity.
[4] For recent overview of differing approaches within simulationism, see Barlassina and Gordon 2017. For discussion of “re-use” in particular, see e.g. Hurley 2008.
[5] Consider the case of i-desire (see Blumberg and Strohminger 2024 and references therein for discussion).
References
Barlassina, Luca, and Gordon, Robert. 2017. “Folk Psychology as Mental Simulation.” Stanford Encyclopedia of Philosophy (Summer 2017 edition), ed. E. N. Zalta. URL= < https://plato.stanford.edu/cgi-bin/encyclopedia/archinfo.cgi?entry=folkpsych-simulation>.
Blumberg, Kyle, and Strohminger, Margot. 2024. Imaginative Hopes and Other Desires. Forthcoming in Analysis. Preprint available on PhilPapers. URL= < https://philpapers.org/rec/BLUIHA>.
Hopkins, Robert. 2024. The Profile of Imagining. Oxford: Oxford University Press.
Hume, David. 1975 (1739). A Treatise of Human Nature, ed. L.A. Selby-Bigge, 2nd ed. Revised by P.H. Nidditch. Oxford: Clarendon Press.
Hurley, Susan. 2008. “Understanding Simulation.” Philosophy and Phenomenological Research 77:3, 755-74.
Kind, Amy. 2017. “Imaginative Vividness.” Journal of the American Philosophical Association 3:1, 32-50.
Langkau, Julia. 2021. “Two Kinds of Imaginative Vividness.” Canadian Journal of Philosophy 51:1, 33-47.
Reply to Strohminger
“If then one said: ‘Images are inner pictures, resembling or exactly like my visual impressions, only subject to my will’ - the first thing is that this doesn’t yet make sense.
For if someone has learnt to report what he sees over there, or what seems to him to be over there, it surely isn’t clear to him what it would mean if he were ordered now to see this over there, or now to have this seem to him to be over there.” (Wittgenstein 1958: §642)
Margot Strohminger’s simulationist tells us that imagining is a way of simulating other mental states and mental processes. (I’ll concentrate here on sensory imagining and the perceptual states it supposedly simulates.) That character further tells us that imagining resembles those other states and processes in certain key respects, and that it is in virtue of these resemblances that it is able to simulate them. Of course, the simulationist will also want to concede some important differences between imaginings and the states simulated. Can’t, Strohminger asks, the Profiling account be seen as a way of putting flesh on these bones?
The short answer, as indicated in the quotation from the book in her note 2, is Yes. Even at its most schematic, simulationism is not quite empty: the postulation of unspecified resemblances and differences is given some content by supposing that, whatever they are, they serve to explain imagining's simulating role (whatever that may be). And, though they take this opportunity far less often than they ought, simulationists are free to fill out this very thin position by specifying similarities and differences. One way to do that would be by appeal to Profiling.
Still, in considering this eirenic proposal we should proceed with caution. The Profiling account is motivated by the desire to do justice to the differences between sensory imagining and perceiving (and between experiential imagining more generally and the corresponding experiences). In part, this is because those differences tend to be underplayed by simulationists. In part, it is because, as Wittgenstein reminds us, differences have consequences. Not every package of putative points of resemblance and contrast is coherent. The book seeks to attend carefully to the many and various differences, and to work out their consequences. The four phenomenological features Strohminger cites are only the beginning of the differences it describes. (For reviews, of features both phenomenological and functional, see pp.142-145 and 251-254.) To cite just two others, I argue that, unlike perception, imagining does not furnish a field of objects across which attention can roam (pp.132-141); and, crucially, that imagining is given to us as representational, but perception is not (ch.2).
Nonetheless, as Strohminger notes, a core of commonality remains. In sensory imagining, just as in perception, non-perspectival properties such as 3D shape and surface colour are present to us as patterns (‘profiles’) in perspectival appearances. This commonality can be used to explain, inter alia, how we are able to use imagining to learn how things will go in the world we perceive (ch.7). Is this not, in effect, simulationism in action?
Perhaps it is. Even here, though, the closer we look, the more differences we find. The workings of profiles in perception exploit the structure of the perceived world. As one walks round a boulder, the visible figure it presents is determined by two facts independent of one’s mental state: the boulder’s shape and one’s spatial relation to it. When one imagines walking round it, matters are quite otherwise. Here, it’s not just that something determines the visible figure one conjures up, only that determination is done by other aspects of one’s mental condition. Rather, one imagines a boulder of that shape only in virtue of the fact that one conjures the appropriate visible figure, for any given imagined spatial relation to it. The perspectival determines the non-perspectival, not the converse. The simulationist may want to describe this as imagining ‘using or reusing the mechanisms of visual processing’, but she should do so only if that inclination survives acknowledging how deep the differences between the pair go.
All that said, I completely agree with Strohminger that there is no inconsistency between the Profiling account and the simulationist program. (Can a theory and a program even be inconsistent?) She floats the idea that the program has generality on its side, but is appropriately cautious about whether this really is to its credit. As the sage might have said, he who says nothing speaks with equal truth of all. Simulationists don’t say quite nothing, but the invitation remains open for them to say enough to stake a real position.
References
Ludwig Wittgenstein 1958 Blue and Brown Books Oxford: Blackwell.