I’m happy to kick off our latest Mind & Language symposium on Bence Nanay’s “The Role of Imagination in Decision-Making,” from the journal’s February 2016 issue, with commentaries by Amy Kind (Claremont McKenna ) and Neil Van Leeuwen (Georgia State).
The psychological process of decision-making is often explained in terms of an agent’s beliefs and desires (or other pro-attitudes): practical reasoning is essentially a matter of estimating the probability of desire-satisfaction for each action in a set of possible actions in light of relevant background beliefs. In his article, Bence points to a number of empirical findings about how human beings actually make decisions that seem to conflict with the belief-desire model. He also argues that the belief-desire model overintellectualizes the decision-making process: “Animals and preverbal infants perceive and perform actions. Should we assume then that each time they perform a perceptually guided action, they go through the appropriate beliefs and desires? Further, arguably, most of our own actions are also largely akin to those of animals: when we tie our shoelaces or brush our teeth, these actions are also unlikely to be mediated by propositional attitudes like beliefs and desires” (127).
Bence’s aim is to outline an alternative model that is consistent with empirical findings and that does not attach theoretical priority to beliefs and desires. Instead, his model emphasizes the role that imagination plays in practical reasoning: when we decide between two possible actions, we imagine ourselves in the situation that we imagine to be the outcome of these two actions and then compare these two imaginings.
Comments on this post will be open for at least a couple of weeks. Many thanks to Bence, Amy, and Neil. All of us here at the Brains Blog are grateful to Sam Guttenplan, the other Mind & Language editors, and the staff at Wiley-Blackwell (especially Carmen Sherry) for their continued support of these symposia!
You can learn more about Bence and his research here.
Below there are links to a video introduction, the target article, commentaries, and Bence’s replies.
Target Article: Bence Nanay, “The Role of Imagination in Decision-Making“
Bence’s video introduction:
Commentaries and Replies
- Amy Kind, “Unconscious Imagination and the Imagination-Model of Decision-Making“
- Neil Van Leeuwen, “Partially Doxastic Imaginings“
- Bence Nanay, Replies to Kind and Van Leeuwen
Hi Bence,
I’m *very* sympathetic to the idea that the kind of imagining you talk about in the target article plays a major role in decision-making (and also to the idea that the belief-desire model over-intellectualizes practical reasoning), however, I’m also inclined to think that decision-making isn’t a psychological kind, but instead a pretty disunified collection of psychological processes. Some utilize the perceptual/imaginative and motoric machinery that we share with many non-human animals; some involve propositionally articulated, belief-and-desire-based practical reasoning; and some are a complicated blend. Here are a few examples. How would the imagination-model handle them?
a) I’m between sessions at a philosophy of conference and feeling sleepy. I would *love* to have some coffee, but am not sure whether I have enough time. The next session, which I really want to attend, is due to start in 10 minutes. I begin to ask around: where can I find some coffee? I am told that the nearest café is about five minutes away. That means that with the to and fro I’d certainly miss the first few minutes of the next session. So I decide to stay put and have a glass of cold water instead. It doesn’t seem like I need to imagine different situations and then compare my imaginings in this example (although I could). The belief-desire model seems to fit reasonably well here.
b) What about epistemic actions in decision-making? To use the famous example, when Tetris players need to decide on the horizontal positioning of a zoid, they can either rotate the zoid imaginatively (make-perceive or mental animation) or “physically” rotate the zoid on the screen. A lot of real-life problem-solving and decision-making arguably also involves deciding which action to perform by selectively modifying the world in different respects so that the right action “pops out” at us. This would be a case where neither the belief-desire model nor your imagination model fits.
c) Many decisions are made with others. I’m between sessions again at a conference and planning lunch with a friend. What are the options? Well, there’s a nice Thai restaurant, a so-so taqueria, and a diner, we are told. I had Thai last night, my friend doesn’t like diner food, and so we settle on the so-so taqueria. How does imagination enter into this scenario?
d) And, finally, consider the case of Chrysippus’s dog. Does the dog need to imagine the consequences of going down the unsniffed path as well as the consequences of not going down the unsniffed path in order to select it?
Just a quick question for Bence; I’m still not quite clear after having (quickly) read the articles and commentaries how low you’re willing to go with imagination. You mention nonhuman animals a few times, but it’s pretty implausible that they would deploy imagination in all of the three ways you suggest in your article. To get even one of the three ways, you’d probably need a fairly deflationary conception of the imagination. For example, a case I sometimes think about is a cat “deciding” whether to make a difficult jump, scooting back and forth, wiggling butt, etc. before leaping or not. Perhaps it’s plausible there that the cat is doing something like scanning its visual and proprioceptive input for affordance of a successful jump. Would this process count as “imagining” a jump? If imagination can go that low, then it could probably accomodate quite a lot of flexible behavior in non-linguistic animals, but that looks like a much less sophisticated process than your job choice case. (And if that doesn’t count as a “decision” on your view, what’s the lower bound?)
I think part of the problem here is that (as Robert notes) “decisions” are a pretty diverse lot and any concept made to accommodate all of it will have to be stretched pretty far. (And if we’re willing to stretch imagination that far, why not belief? Your critique is most powerful against more inflated notions of belief and decision theory, whereas more deflationary approaches like Dretske’s would have more to say in response, at least for the lower bound.)
Thanks a lot, Robert, I fully agree that decision-making is not a monolithic category. And I definitely do not intend this imagination-centered account to be a necessary condition for ALL decision-making.
Here is a decision that is definitely not based on comparing imaginative episodes: you’re deciding between buying a large packet of Oreo cookies or two small ones. The large one has 56 cookies for 9.99. The small one has 27 for 4.99. Your decision will not be based on imagining doing this and then doing that. It will be based on beliefs and desires (and maths!).
So the aim here was not to argue that ALL decisions follow this comparing imaginative episodes patterns, but that many of them do and, crucially, many that have been taken to involve beliefs and desires (and most decisions that really matter) do!
So, in the light of this, your examples:
(a) Coffee or being punctual? My very unreliable introspection tells me that imagination does play a role in this decision: I imagine myself waiting in line for the coffee, knowing that the talk has already started, rushing back to the conference trying not to spill my coffee, sneaking into the venue without being too conspicuous, etc. And I imagine myself nodding off at a talk that I really want to follow, etc. However I decide, it really doesn’t seem to me that it’s just ‘cold’ beliefs and desires. This is not much of an argument, merely an unreliable introspective report, but I’d be curious to hear whether anyone can recognise this way of making decisions of this kind. Again, this is not to say that there are no decisions involving only beliefs and desires (see the Oreo example above) – but this one about the quick coffee doesn’t strike me as one.
Thanks for your response, Bence! The scope of the imagination-account wasn’t clear to me when reading the article. I took you be arguing that the relevant type of imagination is necessary for decision-making in general: “this is the crucial point, beliefs and desires are not the only ingredients that are needed for making decisions—we also need an imaginative episode: we need to imagine the two potential outcomes and compare the two” (p. 134).
Yes, I clearly could have been more explicit about this – I was trying to allow space for the pure belief-desire stuff in the very last paragraph, but it would have been good to give something like the Oreo example there…
Example (b) is really interesting because it draws attention to the relation between mental imagery and imagination (something Amy Kind’s commentary touches on as well). It is known that playing tetris engages one’s mental imagery (there are clinical practices of preventing certain mental imagery by tying down the subjects’ mental imagery with playing tetris). So while I don’t want to insist that this kind of decision (say, whether I rotate it to the left or to the right) involves the kind of imaginative episodes I talked about, but it may involve something similar. And this is something that came up in Amy’s commentary as well. Sometimes all that is needed for the decision between doing A and doing B is that the mental imagery of A is unpleasant or that the mental imagery of B is pleasant. This could be thought of as a very bare-bone and rudimentary version of the imagination-based model I talk about in the paper.
Just saw your response to the coffee example. I think that it certainly could go the way you describe. But I was thinking of the most relevant factor as mathematical : five minutes each way, ten minutes altogether (not even including a wait in line) = not enough time. A lot like your Oreo example.
Hi Cameron, thanks a lot for your comment. What I would say about decision-making of non-human animals (which we actually do know a fair bit about) is something along the lines of my response to Robert. So in the case of the decision-making of non-human animals (and man of our own decisions) all that is needed for the decision between doing A and doing B is that the mental imagery of A is unpleasant or that the mental imagery of B is pleasant. And we know for sure that mammals and birds have mental imagery.
And I guess that would be my way of describing the notorious Chrysippus’s dog example (and not with the help of disjunctive syllogism…)
Again, this would be a rudimentary version of the imagination-based model. But to be frank a lot of this would depend on how we think about the relation between imagination and mental imagery (something I’m blatantly bypassing here…)
Re: the Tetris example, Kirsh & Maglio 1994 (https://onlinelibrary.wiley.com/doi/10.1207/s15516709cog1804_1/abstract) found that inexpert players used an internal, mental rotation strategy, but that experts players “offloaded” their computations onto the external world (here the virtual world of Tetris). This is supposed to be an example of “epistemic action” because it involves changing the environment to make a problem easier to solve rather than changing internal representations of the environment.
Hi Robert, not a topic for here, but I’m extremely sceptical of the Kirsh&Maglio explanation (which is highly (4E) theory-driven, of course).
But the concept of epistemic action is surely important, 4E approaches aside, and seems to explain certain intellectually non-sophisticated kinds of decision-making. Anyway, it’s enough that some people *could* play Tetris using the epistemic action strategy.
Hi Bence, I really enjoyed your target article as well as the replies by Amy and Neil. It has given me a lot to think about. :-))
I completely agree with your claim that the belief-desire model is inadequate and imagination needs to be given a central place in models of decision-making.
You got to this conclusion by thinking carefully about the kinds of characteristic errors in decision-making documented by psychology researchers. This is a great argument and, while I am still chewing it over, I think it is very promising.
I got to a very similar conclusion by a different route: the affective redeployment hypothesis (my forthcoming chapter, called “Deliberative Guidance”, is on my website). The basic idea is that affect is the brain’s primary (only?) currency for registering the value of states of affairs. But affective systems are built to operate on percepts (you see a snake and feel fear, see a friend and feel happy, etc.). To get affective systems to work in deliberation for evaluating *prospective* states of affairs, you have to supply them with the kinds of inputs they expect, i.e., perception-like representations. So you end up with the view that imagining is central to deliberative decision-making – it is needed to construct the sort of representations that can engage affective systems.
Your argument and my argument are almost entirely independent (I think). Nice to see that they seem to be pointing at the same conclusion!
Hi Chandra, thanks a lot. I’m looking forward to reading your ‘Deliberative guidance’ – sounds very compatible with what I’m trying to do here. Also, these affectively laden perception-like representations you talk about sound exactly like mental imagery to me (which is great for my purposes, see my Brains Blog gig next week on mental imagery).
Bence, great paper! I have two questions for you. (i) I like Amy’s example a lot. How open are you to having a “softer” view of prospective reasoning, where the prospection is thought of broadly, in the sense that it could involve imagination but also a kind of cognitive modeling of possible situations that might not, strictly speaking, count as imagining? You could embrace Amy’s example but allow that a decision could involve a kind of prospective assessment that might not amount to an act of imagination (whether conscious or unconscious), yet this type of assessment still involves an interesting sort of prospection that the belief model doesn’t accommodate. (I like this view, since I like to think in terms of prospective cognitive modeling in a way that includes, but doesn’t require, conscious imagining.) So I’m very friendly to your account, but I want to defend the importance of prospective assessment in this broader sense, where imagination *along with other types of epistemic abilities* can be exploited in order to cognitively model and assess possible situations. Do you have an objection to this way of proceeding? (ii) What do you think about cases of transformative experience where we lack the ability to imagine possible future outcomes and by extension to imagine our possible future selves? How should we handle big life decision making in that context?
All this sounds right to me, Laurie. My impression is that when psychology folks talk about “imagination,” they often have this sort of prospective assessment (but also different types of counterfactual reasoning about the past) in mind. For example, Ruth Byrne and Jonathan Evans seem use the term “imagination” in this way. Evans has told me that he is not a normativist, so I think that Bence’s view of imagination is actually very close to that of Evans. What do you think, Bence?
Thanks a lot, Laurie, I’d be very open this such a flexible way of thinking about imagination (I don’t think that imagination is a ‘thing’ – mental imagery is a ‘thing’, presumably counterfactual thinking is a ‘thing’, but not imagination). What I wanted to do here is to work out one way in which we make (touchy-feely) decisions and this involves mental processes that map onto fairly standard ways of talking about imagination.
And there are a lot of really interesting links between this way of thinking about decision-making and your research on transformative experiences. I think of the three imaginative episodes that I identified all of them can just fail to latch onto anything meaningful in transformative decisions. You have no idea what you’ll be like, you can only imagine what you’ll be like (when you have kids, say), so here your imagination is going to be completely unsupported by your beliefs (this is something that came up in Neil’s comments and my responses to them).
Thanks, your reply about your goals for the paper makes sense to me. I think your replies to me and to Neil (and Neil’s comments) raise very interesting issues about how imagination and belief might constrain each other (and what can happen when those constraints fail) in prospective assessment.
Hi Robert, yes, I also thought at some point that Evans and I are on the same page, but he says things like this:
‘we need somehow to imagine the world (in relevant respects)
as it might be following a particular choice or action under our control and decide how much we would like to be living in it. Moreover, we need to conduct a set of thought experiments for each possible action and compare their evaluations’ (Evans, 2007, p. 12).
This sounds very normativist to me. And in the Evans and Over 2004 book, they are even more explicit about this.
Hi Bence,
That passage doesn’t sound especially normativist to me. Check out Elqayam and Evans (2011). “Subtracting ‘ought’ from ‘is’: Descriptivism versus normativism in the study of human thinking.” BBS, 34: 233–290; especially section 6.1.
https://journals.cambridge.org/action/displayFulltext?type=1&fid=8405997&jid=BBS&volumeId=34&issueId=05&aid=8405995&bodyId=&membershipNumber=&societyETOCSession=&specialArticle=Y
The whole article is a siege assault on normativism.
Thanks Robert. Maybe normativism is not a very helpful label here as it can mean so many things. I read Evans (in spite of the more general methodological claims of the BBS paper) as being primarily interested in how we should make decisions in order to arrive at the optimal outcome. And I am interested in how we actually make decisions (irrespective of whether the outcome is optimal (and the whole issue about optimal outcome is even more problematic in the light of the transformative decisions Laurie and I were corresponding about above)). But, again, I’d be delighted if Evans’s account turned out to be more similar to my own!!