A post by Anatolii Kozlov

It is said that pretense behaviour requires guiders that can navigate and channel otherwise unconstrained imaginative activity.

For example, according to Shaun Nichols and Stephen Stich, pretense behaviour is guided by the propositional instructions, provided by the Possible World Box, ‘a work space in which our cognitive system builds and temporarily stores representations of one or another possible world’ (Nichols and Stich 2000). The difficulty, however, is that if pretense behaviour is indeed fully sanctioned by these propositional instructions, to supply them there must be an infinite number of conditional beliefs that would define every single aspect of pretense behaviour. Suzanna Rucińska, identifies it as a problem of infinite regress (Rucińska 2014).

In contrast to the propositional view, Neil Van Leeuwen suggests that pretense activity is guided by transparent sensory imagination that can be a part of the visual field: ‘There exists a form of imagining that is a continuously updated forward model of action in the world /…/ A “forward model” is an internal representation of motor commands that anticipates the consequences of those commands on bodily motion. The “nonveridical” perceptual representations are basically mental imagery’ (Van Leeuwen 2011). On a parallel note, in (2014) he defines a constructive imagination as a capacity to form novel mental images out of memory ‘acquired by perceptual or other experiences.’ At the same time, he admits that the concept of constructive imagination faces a problem of under-specification. If in imagining a dancing cat one pictures a cat in a tutu and not in a black hat, it is not clear why specifically such an image occurred in imagination.

Now, if we disregard the exact format of guiding representations, both the infinite regress and under-specification problems seem to be two sides of the same coin. If pretense behaviour is guided by some mental representations, the question is how exactly those mental representations are selected and narrowed down, given the number of alternatives that are possible?

Rucińska sidesteps the problem of constructing internal representations by externalising pretense. She suggests that pretense behaviour can be seen as explorative rather than purely intentional and she identifies environmental and social affordances as the sources of pretense guiders.

In a general sense, affordances can be understood as possibilities for action. For example, bananas afford to us holding them as if they are telephones. In a similar way, if there are other participants in the pretense game that we are playing, they provide us with the possibility to respond to their reactions. Thus, in pretending something, we don’t have to come up with some internal representations of the premise but simply respond to the existing affordances.

I am generally sympathetic with the idea of external affordances as the source of pretense guiders. However, I think that the explanation of pretense behaviour still must have a place for internal representations. As an example in favour of this, I’m thinking about the case of pretense in the form of pantomime done without external props. Take, for instance, the piece by Rowan Atkinson called ‘The Piano Player’:

Using nothing except his own body, the comedian enacts a piano performance, intricate in both technical and emotional senses. Of course, by the time of this recording, the sketch must have been rehearsed to the highest level of bodily automatism, but it is still hard to see how it can be performed – or at least initially developed – without guidance by any internal representation, e.g., a (quasi)image of a piano and piano keyboard. In pretending, we do sometimes rely on sensory representations because they can provide affordances in the absence of external cues. Moreover, the more nuanced the image, the more specific affordances it would provide.

But then we are back again to the problem of multiple possibilities of mental representations. In enacting a certain pretense scenario, we might rely on non-arbitrary mental representations, which are specific variants singled out of many possible mental representations. In Nichols and Stich’s account, the variant is singled out through exhaustive prescriptions generated in the Possible World Box (supplemented by the Script Elaborator). In Van Leeuwen’s account, the representation is a product of recombination between different perceptual (and other) experiences, which is further constrained by contextually relevant beliefs.

In my interpretation, imaginative mental representations are a ‘quarantined’ product of the ‘clash’ between groups of beliefs (and background knowledge) associated with each of the elements of the imaginative premise. By ‘quarantined’ I mean that the imaginings do not affect the body of my beliefs. By ‘clash’ I mean a convoluted procedure, which includes retrieval of my beliefs about each element of the premise and comparison of those beliefs throughout the different relevant dimensions, which allows delineating the scope of possibilities for the imagined representations. This procedure, as I suggest, is the most efficiently done if mental imagery is involved.

Let’s consider the following example: I know that I am not a rabbit, but I can pretend that I am. On the one hand, I can immediately act out my pretense, that is, perform whatever relevant actions come to my mind here and now. On the other hand, I can stretch this exercise over a long period of time: instead of acting like a rabbit at this very moment, I can generate some imaginings of a ‘rabbit-like’ me and keep them in my memory until the later moment when I would act them out.

My first point here would be that to produce ‘rabbit-like me’ imaginings, minimally, I need to have beliefs about myself {B1} and beliefs about rabbits {B2}.

My second point is that my resulting imaginings of a ‘rabbit-like me’ {B1;B2} are going to be some kind of conjunction and/or intersection of my beliefs about myself {B1} and beliefs about rabbits {B2}. And even if they are not such, my beliefs {B1} and {B2} are going to affect my imaginings by providing a point of reference.

For example, in thinking about the heights of rabbits I believe that rabbits (generally) are low. In thinking about my own height, I have a pretty solid idea of what my height is (certainly I’m higher than an average rabbit). But in thinking about the height of a ‘rabbit-like me’, I am presented with a spectrum of possibilities. In a way, I am constrained by my beliefs about the height of rabbits and my own height. These beliefs provide a point of reference: unless my game specifies otherwise, the height of a ‘rabbit-like’ me would likely be somewhere between the height of rabbits and my own height.

The problem is that retrieving beliefs is generally a simple issue.  But deciding the height of the imagined creature is not a simple issue. What I need to do (implicitly) is to retrieve beliefs about the height {B1}, beliefs about the height {B2}, and through their comparison evaluate the spectrum of possibilities that they provide.

The difficulty here is that the height is only one dimension according to which I can compare myself to rabbits, while in fact there is a multitude of such dimensions. The same way I am presented with the spectrum of possibilities within one dimension, I am also presented with the spectrum of possibilities regarding the dimensions of comparison themselves. It can be the height, but it can be the ear shape, behaviour in one or another situation, and many other things. The more dimensions I would address, the more developed my pretense would be.

In summary, to imagine {B1; B2} would be akin to scanning through the dimensions of comparison between {B1} and {B2} and identifying the range of ‘values’ for each of the dimensions that appear to be relevant within the given imaginative task.

Although this way of accounting for imagination doesn’t give a full explanation of how the exact mental representations are produced, it does help to see how beliefs narrow down the scope of mental representations. Besides, it provides a view of why the construction of mental representations, even propositional ones, is in dire need of some ‘technical’ mental imagery. If constructive imagination indeed involves coordinated assessment of clusters of beliefs, this procedure can be facilitated if performed on the entities with map-like properties, and mental images have them.

References 

Nichols, S, and Stich. 2000. ‘A Cognitive Theory of Pretense’. Cognition 74 (2): 115–47. https://doi.org/10.1016/S0010-0277(99)00070-0. 

Rucińska, Zuzanna. 2016. ‘What Guides Pretence? Towards the Interactive and the Narrative Approaches’. Phenomenology and the Cognitive Sciences 15 (1): 117–33. https://doi.org/10.1007/s11097-014-9381-z.

Van Leeuwen, Neil. 2013. ‘The Meanings of “Imagine” Part I: Constructive Imagination’. Philosophy Compass 8 (3): 220–30. https://doi.org/10.1111/j.1747-9991.2012.00508.x.