week 8: MENTAL IMAGERY 8A: MENTAL IMAGES There is a certain phenomenology (i.e., a certain way the world appears to the experiencer) that doesn’t seem to be readily analyzed in terms of conceptual or propositional content, no matter how elaborate Mental images are thought to be similar to external images (photographs, paintings, films) in the sense that they also have a kind of phenomenological richness that we could never fully capture by trying to translate them LONG HISTORY OF A DEBATE In fact, Plato thought mental images were doubly misleading, since, in his cognitive ontology, images are copies of copies For Plato, then, if you fail to grasp the underlying concept, and think instead in purely imagistic ways, you will fail to gain knowledge. Your cognition will be degenerate in some way However, Berkeley argued that no idea of a triangle could contain these contents alone since, for him, every representation had to be of a particular experience or finite set of particular experiences In reality we have a number of quite different particular experiences of triangles stored in memory that are available for recall Thus, Berkeley provided us with an early exemplar theory of concepts, believing it to be the only theory of concepts that could explain how we derive contents from the imagistic representations One of the founders of scientific psychology, Wilhelm Wundt, agreed with David Hume that there are really no differences (from the point of view of the experiencing subject) between representations arising from perceptual experience and those arising from memory If visualization works by accessing visual memory, then other sense modalities can produce their images too ARE MENTAL IMAGES REALLY LIKE PERCEPTS All of the fine-grained descriptive information had to be added sequentially in order to answer the increasingly detailed questions that were posed So perhaps Berkeley was wrong to think that mental images must always contain all the particularized detail of ordinary perceptual experiences A PESSIMIST WORRY Is the idea of a mental “image” just a metaphor? Metaphors are great because of the ways they can structure and inform early theorizing about a topic Those who argue for this view — for example, Daniel Dennett — point to all kinds of features we think are stable when it comes to our perception of the world but show that there is actually lots of interesting information loss as we try to construct a coherent scene Change blindness occurs when there is a change in perceptual stimulus but the observer does not notice it. Inattentional blindness is failure to notice prominent perceptual stimuli because our attention is on some other aspects of the scene 8 B: PERCPETION, CONGITION, AND IMAGES Before we speak in further detail about mental imagery, it would be useful to make two distinctions that have only hitherto been implicit in our discussions. The first is the distinction between perception and cognition (as types of mental process) and the second is a distinction between percepts and concepts (as types of mental representation) PERCPETION: COSNTRUCTING A SCENE Perception is a special kind of information processing by which the perceiver recognizes and “interprets” sensory stimulation The coherent and unified view of self and world comes together in the percept, which is a mental representation developed as a causal consequence of the perceptual process Kant argued that if experience is going to be possible at all, then the chaos of data flowing in from our sensory organs must be combined in such a way that the percept is unified before consciousness Kant thought that this unity of experience implied a unity of self — what he called “the transcendental unity of apperception” concepts must be operating within perception in order to (a) give unity to percepts, and (b) provide percepts with some of their intentional content The unity of the percept is constructed out of tiny packets of information that are the product of rapid eye movements called saccades PERCEPTION: UNCONSCIOUS INFERENCE AND PREDICTION One way of thinking about the brain is as a big predictive (and interpretive) engine On this model, the brain is a set of layers forming a hierarchy perception is about inferring the world based on our present understanding of it, and changing that understanding only when we get an “error signal” — essentially, some new and surprising sensory input that fails to match up with our stable picture of what’s going on in the world CONCEPTS IN PERCEPTIO Do we build up concepts out of perceptual contents, as folks like Prinz argue? If so, then we can see how concepts might be contributing to perception, since concepts would have a perceptual basis. But this really just shows us that percept and concept share a format It does not tell us how perceptual information stored in memory plays a role in future acts of perception So Prinz is off to a great start, but the story is not yet complete. (He offers a theory of how cognition penetrates perception, but we will not look at the details of that theory here.) We have the capacity for seeing as that seemingly trades in conceptual content. Can we do this without thinking of concepts as having a perceptual basis? Can we do this without mental images? SEEING AND ‘SEEING AS’ Images in the world can be ambiguous; mental images cannot. They are what mediates one definite interpretation over another Mental images account for how we can see the external image as a rabbit or as a duck PRIMING AND SEEING IN … OR RATHER “HEARING IT” Thankfully, we can model perception as admitting of top-down effects Unlike the Muller-Lyer illusion or the McGurk effect, this case of priming is showing us that there can be a great deal of cognitive penetration in perception This means that mental images are alive in at least some normal perception Mental images seem to be operating even in cases of “normal” perception where there is an external stimulus. They seem to be what mediates our interpretation of what we see 8 C: A THEORY OF MENTAL IMAGERY In the first part of this lesson, we will look at two sources of evidence for picture theory: the theory that mental images are like pictures presented to us in a spatial manifold When given prompts before engaging in memory tasks, it turns out imagistic and spatial strategies for remembering increase our ability to recall information Are the like pictures before the mind in an analog space, capable of mental “manipulation” in ways similar to the way we manipulate actual objects in actual space? Or, is there a problem with the theory of mental images as being, literally, image or picture-like? On the functionalist view, the mind’s mental processing is supposed to be like the digital computer’s information processing when it runs a program But if having mental images is something the mind can do, and if mental images are analog rather than discrete, it becomes unclear how we can sustain the comparison between minds and digital computers MENTAL IMAGES AND MEMORY One of your readings this week comes from Pylyshyn. Pylyshyn is perhaps the most famous critic of images-as-pictures accounts of mental imagery, and Paivio is often taken to be one of the main proponents of the idea that (a) images help in certain cognitive tasks (like memory tasks) on verbal contents (words), and (b) that even purely verbal processing rests on imagistic faculties, which we can measure when we see how much easier it is for subjects to reason with nouns that have a high image-value (concrete nouns like “dog”) as opposed to those with a low image-value (e.g. abstract nouns like “justice”) MENTAL ROTATION, MENTAL SCANNING, ANDMENTAL “SPACE” Mental images were seen as being like perception in this way: they shared a common structure. They seemed to be analog images, and these images were presented to us in a space structured very much like the spatial manifold of perceptual experience — continuous, uniform, homogeneous, and isotropic In mental scanning tasks, there is a delay in response time when subjects are asked to make judgments about details of a mental representation that, if picture-like, would be further away from each other in the space occupied by the image Kosslyn has actually argued that picture theories of mental imagery are not synonymous with “analog images” DESCRIPTION THEORY Either mental images are really analog pictures before the mind, calling into question the analogy between mind and computer that governs computational theories in cognitive science, or the mental “pictures” really can be represented in description form This speaks to the descriptive nature of mental images; it’s only by our adding certain varieties of descriptive content that our mental image starts to have the richness of detail that a picture would normally possess If our mental images were like pictures, we would assume that such details would be there to be discovered (as they are in picture) instead of there to be constructed (using more descriptive content) A NOTE ON UNDERDETERMINATION OF THEORIES Another possibility is that we are dealing with two theories that are empirically equivalent As a matter of logic, for any given scientific theory, there exists at least one other incompatible scientific theory that is equally supported by the evidence Perhaps the description-picture debate is an example of such underdetermination. As of now, both accounts do a pretty good job of modelling certain observable phenomena we are interested in, although, as we have seen, there are differences in the kinds of phenomona that each theory best models