M. Jimena Clavel Vázquez is a visiting lecturer at the Philosophy Department of Tartu Ülikool (the University of Tartu), in Estonia. She is interested in situated approaches to cognition, in their notion of embodiment, and in what they can tell us about perception and imagination.
A post by M. Jimena Clavel Vázquez and Adriana Clavel-Vázquez
In The Living Mountain, Nan Shepherd notes that her eyes give her access to many worlds: to light, colour, shape, and shadow, to the patterned world of the snowflake and the petal, and to the rhythmic lines of the mountain. Her eyes, ultimately, give her access to beauty. What enables this access? Shepherd suggests that “[p]erhaps the eye imposes its own rhythm on what is only a confusion”, that “one has to look creatively to see this (…) as beauty” (Shepherd 2014, p. 101). For Shepherd, beauty is not only something we see, but something we see when we look creatively.
Here, we outline a defence of aesthetic perceptualism, the view that aesthetic properties feature in perceptual experience, on the basis of an anticipatory approach to perception.[1] Aesthetic perceptualism faces challenges from the get-go. Typically, it’s said that only low-level properties (like colour and shape for vision, or volume and pitch for sound) feature in the content of perceptual experience. Aesthetic properties are high-level and don’t belong to features paradigmatically associated with sense modalities.
Adriana Clavel-Vázquez is a British Academy Postdoctoral Fellow and Junior Research Fellow at Wolfson College, University of Oxford. Her research focuses on the role imagination plays in our social interactions and engagement with art, and its consequences for the interaction of ethical and aesthetic values.
There is, however, at least one powerful reason to hold on to aesthetic perceptualism even if problematic: it seems to do justice to how we experience aesthetic objects. Regarding the mountains as serene rather than violent feels different. Aesthetic perceptualism explains this phenomenal difference by arguing that aesthetic properties feature in perceptual experience. The challenge for aesthetic perceptualism is to show not only that some aesthetic properties feature in perceptual experience, but that the events that give access to them are on a par with those that give access to low-level properties (Reiland 2014).
Aesthetic properties are good candidates for perceptualism. First, they are gestalt-like in that they arise when low-level properties are perceived organised in specific ways, but are irreducible to these. Second, we can identify in aesthetic properties a scaffolded structure. Low-level properties of objects, like colours or shapes, ground base-line substantive aesthetic properties, like dynamism. These, in turn, ground more complex substantive aesthetic properties that might include an evaluative component, like ebullience. And these, in turn, ground verdictive aesthetic properties, like beauty, which are purely evaluative and concern overall aesthetic merit. Third, base-line substantive aesthetic properties are not primarily aesthetic terms, but terms that have primary non-aesthetic uses.
We think anticipatory approaches to perception can come in handy in explaining how low-level features are perceived organized so that aesthetic properties are brought into perceptual presence. In particular, we defend a version of the sensorimotor theory of perception (e.g., Noë 2004; O’Regan & Noë 2001) that draws on predictive processing (e.g., Clark 2016; Hohwy 2013).
Proponents of the sensorimotor theory aim to explain that what enjoys perceptual presence doesn’t correspond entirely with what impinges our senses at a time. The content and phenomenal character of perceptual experience is due to perceivers’ possession of practical knowledge of how sensory information would change after movements, either of the perceiver or of the object perceived. Consider an unseen aspect of an object, like the back of a bottle. For the sensorimotor view, the back of a bottle makes a difference to our perceptual experience—it is present as absent—because it is virtually available to the perceiver: she could look around and get a glimpse of its back.
The sensorimotor view provides an interesting story about the way certain properties come to feature in our perceptual experience. According to the dual content view (Noë 2004), perceptual content involves two aspects: objects’ appearance properties (p-properties)—which change continuously and depend on the specific location, angle, and time of the perceptual episode—and objects’ intrinsic properties. What enables going from p-properties to intrinsic properties are expectations of the way sensory stimuli would change after certain interactions. For instance, we come to see the circular shape of a coin because it is the kind of thing that, if one were to move in certain way, it would reflect an ellipsis in the retina.
Regardless, one might think that sensorimotor expectations are too limited to support the high-level properties in which we are interested. Enter predictive processing.
Predictive processing takes perception to be driven by top-down processing. Predictions about the likely cause of stimuli are advanced based on a generative model. This model can be ‘deepened’ or enriched over time as perceivers interact with their surroundings. Predictions are then compared against incoming stimuli. Additionally, the perceptual system assesses the precision of error feedback obtained when comparing prediction and stimuli, i.e., precision weighting. Based on this process, the generative model is enriched providing, thus, better predictions over time. Perceptual processing is hierarchically structured. At the lowest-level we find predictions about sensory stimuli (e.g., sensorimotor expectations), while at a higher-level we find top-down processes that concern more abstract predictions.
Note that, even when the properties that feature in perceptual experience are due to expectations that often arise from higher, more abstract levels, perceptual experience does not cease to be constrained by current stimuli. Of course, not all expectations are relevant, but only those that are linked to the way present stimuli can be modified continuously through interactions.
Generative models have two features that are relevant for our purposes. Firstly, they involve information from several sources: different sense modalities, proprioceptive and interoceptive information, and significantly for the aesthetic case, affective responses. Secondly, these models are agent-dependent, since they include perceivers as part of the world (Wilkinson 2019).
We believe that marrying the sensorimotor theory with predictive processing makes perceptualism more plausible. The anticipatory mechanisms that explain that low-level properties, like shape or volume, feature in perceptual experience can explain that some high-level properties do, too. This is the case of the non-aesthetic counterparts of substantive properties, like dynamism, unity or balance. It is plausible that these come to feature in perceptual experience as a result of predictions that structure the perception of low-level features on the basis of sensorimotor expectations, other sense modalities, and proprioceptive and interoceptive information.
In some cases, nevertheless, expectations relevant to current stimuli might be insufficient to account for the richness of experience. This is the case of aesthetic objects. When talking about the beauty of the mountains, Shepherd remarks on their fluidity and rhythm, which would not seem pertinent for a static mass of rocks in non-aesthetic contexts. What explains the organization of low-level properties in this case? It seems that sensory imaginings related to, e.g., movement might be playing a role.
To make sense of the precise contribution of sensory imaginings, we need to turn back to the notion of precision weighting (Jones & Wilkinson 2020). Recall that the predictive system processes both the difference between current stimuli and the prediction advanced, as well as the precision of error feedback. Precision weighting is also associated with the allocation of attention to what is deemed to be more reliable (Ransom et al. 2020). Sensory imaginings contribute to precision weighting, firstly, by bringing attention to specific aspects of the object. Second, they make salient certain expectations that were not relevant before, or that don’t normally pertain to the object. In addition to this, sensory imaginings might trigger the deployment of interoceptive and affective predictions prompted by associations to sensory stimuli.
Our view is that in aesthetic cases, the perceptual exploration of an object is driven by a pursuit to fulfil expectations directed by imaginings. Aesthetic perception is driven not so much by what is there, but by what the perceiver can creatively do with what is there. Sensory imaginings drive expectations that organize low-level properties, bringing aesthetic properties into perceptual presence. Just like non-aesthetic dynamism can feature in perceptual experience in virtue of sensorimotor expectations related to movement, aesthetic dynamism can feature in perceptual experience in virtue of expectations driven by sensory imaginings related to movement. As said before, however, sensorimotor expectations are not enough. The growingly complex expectations allowed by predictive processing allow us to move from base-line substantive properties to more complex substantive and verdictive aesthetic properties. Interoceptive and affective predictions driven by sensory imaginings organize low-level properties not only according to, e.g., movement, but to rhythmic and serene movement.
Let’s go back to Shepherd’s remarks. She not only sees the sharp angles and lines of the mountain’s face, but its fluidity and rhythm. Her familiarity with the mountain gives rise to expectations concerning tranquillity. What for others might appear as violent and tortured, for Shepherd appears “as more than jag and pinnacle.” It appears “as beauty.”
[1] The view doesn’t concern artistic properties, among which we can include representational, semantic, symbolic, moral, and aesthetic properties.
References
Clark, A. (2016). Surfing Uncertainty: Prediction, Action, and the Embodied Mind, Oxford, UK: Oxford University Press.
Hohwy, J. (2013). The predictive mind. Oxford, UK: Oxford University Press.
Jones, M., & Wilkinson, S. (2020). From Prediction to Imagination. In The Cambridge Handbook of the Imagination (pp. 94–110).
Noë, A. (2004). Action in Perception. Cambridge, Massachusetts; London, England: The MIT Press.
O’Regan, J. K., & Noë, A. (2001). A sensorimotor account of vision and visual consciousness. Behavioral and Brain Sciences, 24, 939–1031.
Ransom, M., Fazelpour, S., Markovic, J., Kryklywy, J., Thompson, E. T., & Todd, R. M. (2020a). Affect-biased attention and predictive processing. Cognition, 203.
Reiland, I. (2014). On Experiencing High-Level Properties. American Philosophical Quarterly, 51(3), 177–187.
Shepherd, N. (2014). The Living Mountain. The Canons. Canongate Books.
Wilkinson, S. (2019). Distinguishing volumetric content from perceptual presence within a predictive processing framework. Phenomenology and the Cognitive Sciences, 791–800.