January 20th, 2020.pdf - The Cog Dog says \u2026 Do you really \u2018see\u2019 me when you look at me Our perceptions are fallible We sometimes see what isn't

January 20th, 2020.pdf - The Cog Dog says u2026 Do you...

This preview shows page 1 out of 54 pages.

You've reached the end of your free preview.

Want to read all 54 pages?

Unformatted text preview: The Cog Dog says … Do you really ‘see’ me when you look at me? Our perceptions are fallible. We sometimes see what isn't there. We are prey to optical illusions. We are error-prone/ Carl Sagan Last class: Sensation and perception ¤ Sensation ¤ Energy from the environment is absorbed by a sensory organ and converted into a neural impulse ¤ Perception ¤ Sensed information is interpreted as meaningful by the brain ¤ Synesthesia ¤ A condition in which the senses mix ¤ The McGurk effect ¤ Illustrates that visual input can direct speech perception Vision also influences taste perception ¤ Visual information changes taste perception ¤ If food coloring is added to orange juice, participants perceive differences in taste from orange juice without food coloring ¤ Foods of unusual colors (blue steak) are perceived as tasting ‘off ’ and made participants feel sick Hoegg & Alba, 2007; Schlosser, 2001 The early visual system ¤ Light waves enter the eye through the cornea, focused onto the retina ¤ Photoreceptors in the retina convert light to electrical activity ¤ An electrical signal is sent to bipolar cells then ganglion cells ¤ The signal exits the brain through the optic nerve The late visual system ¤ Involved in visual consciousness ¤ Demonstrates functional specialization: modules ¤ There is a retinotopic map in the primary visual cortex Retinatopic map → How information is presented to retina is how it will be processed by visual cortex Visual pathways ¤ The where pathway and what pathway process spatial location and object information, respectively Selective deficits to particular types of objects, depending on where in the brain there is injury ¤ Visual agnosia: Damage to the areas along these visual processing pathways result in selective Person CAN see, just does problems visual recognition not recognize or interpret the visual information ¤ Visual sensory system is intact ¤ Other forms of recognition are intact ¤ Agnosia can selectively affect recognizing certain categories of objects, depending on which area of the brain is damaged Where pathway → Takes information to parietal lobes, processes spatial locations of objects in relation to the person perceiving What pathway → Takes information to temporal lobes for processing, object recognition Visual Modules in the brain ¤ Extrastriate body area (EBA): the brain area involved in processing non-facial body parts. ¤ Parahippocampal place area (PPA): the brain area responsible for the conscious recognition of places and scenes ¤ Fusiform face area (FFA): the brain area responsible for the conscious recognition of faces. Visual Modules: Prosopagnosia ¤ Fusiform face area damage leads to a selective deficit in recognizing faces ¤ The ability to recognize other objects visually is still present Prosopagnosia ¤ Rely On non-facial characteristics to identify people They CAN, however, extract expression from faces and determine whether male or female ¤ Results from brain injury to the FFA, but can also be congenital (genetic) ¤ Sub-types “Face blindness” ¤ A problem perceiving faces Faces look distorted ¤ A problem attaching meaning to faces Cannot identify people based on faces alone Apperceptive visual agnosia ¤ A failure recognizing objects due to problems with perceiving objects ¤ Sensory processing is still intact Can’t bind individual features, although detected, into a meaningful combined percept ¤ Visual features can be detected, but there is problems grouping visual features into meaningful percepts ¤ Cannot recognize objects, draw, or copy an object even if they have knowledge of that object ¤ Damage to occipital cortex Associative visual agnosia ¤ Inability to associate visual input with meaning Can’t access information about objects and link it to what they see ¤ Occipital-temporal cortex damage ¤ Leads to problems on tests that require: ¤ Drawing objects from memory ¤ Naming objects ¤ Indicating the functions objects ¤ Determining if a visual object is a possible or impossible object You can present impossible objects or creatures, and they will recognize it as “normal” Last class: Visual illusions ¤ Expectations about what we should see or how we think the world works ¤ Illustrates top-down processing effects on perceptions Forming perceptions based on personal knowledge Processing directions ¤ Bottom-up processing: the feed-forward influence of the external environment on the resulting perceptual experience ¤ Visual information from the primary visual cortex propagates down the “what” and “where” pathways ¤ Top-down processing: the feed-back influence of knowledge, expectations, and goals on the perceptual experience ¤ Information from the end of the visual pathways is sent back to the visual cortex via feedback connections For the rest of today … ¤ Object (Pattern) recognition theories ¤ The role of cognitive processes in identifying visual input ¤ An ecological approach to perception ¤ Visual cues in our environment direct perception ¤ Context affects perception ¤ Our surrounding affects vision ¤ Gestalt approaches Visual object recognition ¤ Perception involves processing basic visual elements of an input that are added up for object recognition ¤ Pattern recognition theories ¤ Identifying a pattern in visual input as an instance of a member of a category you hold in memory ¤ Match the pattern in visual input to existing patterns ¤ Incorporates memory ie. “barcode” used to search your memory for what the object is Pattern recognition theories ¤ A short-term (primary) memory trace represents visual patterns used to probe long-term memory traces ¤ The similarity between these two types of traces determines recognition ¤ Question: What is the probe being compared to in the longterm memory? Template matching theory ¤ Match the pattern you see, stored in short-term primary memory (probe) to memory traces (templates of previous objects you have encountered) in long-term (secondary) memory Uses too many cognitive resources - Would require a template for everything you’ve seen ¤ We probably don’t do one-to-one template matching because ¤ We can recognize new objects that don’t match a template ¤ We can recognize objects from different viewpoints But what about this guy you have never seen before Prototype theory most basic features to help ¤ Compare visual pattern with a prototype Compares compare and identify ¤ The average representation of the object in memory formed from what is common among all encountered instances ¤ Allows for ‘flexible’ object identification because a good enough match (resemblance) can support object recognition Prototypes and false memories ¤ Participants studied a series of faces related to a prototype face Each face seen systematically varied from the actual prototype ¤ Did NOT study the prototype ¤ Participants completed a recognition memory test that included the studied faces, new faces and the prototype studying those faces, they created a prototype in their mind as an face Whenaverage of all the faces they had seen, so when shown the actual prototype, they recognize it ¤ People falsely remembered studying the prototype face and accurately remembered the studied faces Solso & McCarthy, 1981 Feature detection theories ¤ Complex input is broken down into individual parts (features) for recognition ie. Bike → Wheels, handle bars, pedals ¤ Features are simple element that can appear in combination with other visual elements to form objects ¤ Each feature processed separately and then assembled into a pattern for recognition ¤ Selfridge’s (1959) pandemonium model Selfridge’s pandemonium model ¤ Bottom level: data or feature level ¤ individual features are represented ¤ properties such as size, colour, shape, etc. ¤ Middle level: cognitive level ¤ detects particular patterns of features and decides if it matches its particular pattern Matches it to the features presented at bottom level ¤ Top level: decision level ¤ decides which pattern is being recognized based on the input of the cognitive demons The one with the greatest similarity is what proceeds through the thinking pipeline Selfridge’s pandemonium model “Decision demon listens to the loudest cognitive demon” If there are two objects that share a lot of similar features, they will be similarly represented at the cognitive level, making it difficult to differentiate at the decision level Feature detection Bottom level of Selfridge’s pandemonium model ¤ Feature detection occurs in the primary visual cortex ¤ Special cells/areas for processing different visual features (color, lines, etc.) ¤ Parallel processing ¤ We performs many computations at once rather than processing information in serial order Recognition by components ¤ All objects are reducible to a set of geons, basic geometric shapes ¤ There are 36 geons that can be combined to form any object ¤ Object recognition involves: ¤ Mentally separate the object into geons ¤ Comparing the geon arrangement with existing memory representations Removes the arbitrary element of having to identify a feature Recognition by components Geon properties ¤ Geons have distinct properties that we can perceive from any angle/view/perspective ¤ Accounts for how we can recognize objects from different viewpoints Object recognition in noise ¤ Geons are resistant to noise and this feature explains why object recognition is robust Fewer Geons: Harder! More Geons: Easier! Image demonstrates a flashlight An ecological approach ¤ Perception results from direct contact of the sensory organs with the environment ¤ There is enough information in our visual environment to perceive ¤ We don’t have to ‘transform’ sensory information and information from the visual environment guides perception ¤ No higher cognitive processes are needed ¤ A passive bottom-up approach J.J. Gibson Perception can only be understood if studied in the REAL world, as it occurs through the direct contact of sensory organs and external environment Gibson’s ecological approach ¤ Perception must be studied in the real world ¤ Cues in the ambient optical array (AOA) direct perception ¤ The AOA is the visual information present at a point of view ¤ the patterns of light reflected from surfaces and objects that change with conditions ¤ What changes in the AOA are the cues about perception ¤ Texture gradients ¤ Topological breakages ¤ Scatter-reflection (i.e., how widely light scatters) 1. Texture gradients ¤ The density of a texture (gradient) provides information about orientation and distance of an object ¤ Incremental changes in the gradient provides information about angle or slant 1. Texture gradients ie. Top of magic kingdom looks higher up than it actually is 2. Topological breakages ¤ Discontinuity created by the intersection of two textures provides information about edges of object ¤ Object identification 3. Scatter reflection ¤ The degree to which light scatters when reflected from a surface, which tells us about the smoothness of a surface ¤ Rough surfaces: Light will scatter more widely ¤ Smooth surface: Light is less scattered Bounces off very cleanly, with fewer angles Gibson’s ecological approach Purely “bottom-up”, never “top-down” ¤ How we perceive from the environment determines actions Cues that are automatically seen, in environment (don’t have to be consciously thinking about it) ¤ Affordances ¤ Visual cues in the AOA ¤ Provides information on the potential function of an object ¤ Buttons, levers, slots ¤ Perceived directly and immediately Intimate relationship between perception and action ¤ Perception and action are linked Opening a door ¤ How you grasp a handle (action) is determined by your perception Animal Face Image is the same, however when told “animal”, mouse is seen, and when told “face”, a face of a man is seen (“topdown” processing) Bugelski and Alampay (1961) Context affects visual perception ¤ Change in perception of a visual element based on the surrounding information and the observer’s perspective and prior knowledge Seeing different things, based on expectations ¤ Top down processing ¤ Three examples ¤ Ames Room ¤ Letter in Context ¤ Color in Context Ames Room ¤ A functional illusion ¤ When expectations are used to perceive ¤ Expectations of observation Slanted floor makes height look significantly different, however we expect that it’s simply a normal room Letters in context ¤ The ability to read words in sentences even when the letters in the middle of some of the words are mixed up ¤ You expect to see real words in a sentence ¤ Letters in correct positions of jumbled words help with This is why it’s so hard to edit our own writing pieces, word recognition because we already have the association that everything are real words You can stlil raed this senetnece even thuogh lettres in the wrods are jubmled. Color in context ¤ The context a colors appears in can influence how you see that color. ¤ Color perception depends on both: ¤ the wavelengths of light that fall on our retina ¤ our past experiences of how objects look under different contexts of illumination The Munker-White Illusion Shading of grey appears different, but they’re actually the same Darker - Black bands on top of A column Lighter - More white around the grey Another example: The Bezold effect ¤ Colors are perceived as a function of adjacent colors The Gestalt approach to perception ¤ The whole that is perceived is greater than the sum of its parts ¤ Holistic rather than piecemeal approach to perception ¤ Top-down processing is what guides perception Bi stable figures What do you see? A Few Gestalt organizing principles To see whole objects, elements must be grouped together in a particular way to be perceived in a particular way 1. Principle of experience 2. Principle of proximity 3. Principle of closed forms 4. Principle of good contour 5. Principle of similarity 6. Principle of common movement 1. Principle of experience 80 COGNITION Elements are segmented based on familiarity with figures ¤ Experience affects the figure-ground segmentation process ¤ Regions usually perceived as the figures are the ones with higher denotivity ¤ those that appear more familiar and easily named to the observer (a) (b) FIGURE 3.19 Bi-stable figures From: (1).gif and hrd/history/gestalt.gif. 2. Principle of proximity (a) (b) FIGURE 3.22 The principles of proximity (a) and closed forms (b) Adapted from: Katz, D. (1951). Gestalt psychology: Its nature and significance. London: Methuen. Objects or features that are close to one another in a scene will be judged as belonging together 3. Principle of closed forms Although containing gaps, our brain “fills in the blanks” to see the object as a closed form We see a shape in terms of closed forms and we like to see items that enclosed as whole COGNITION 4. Principle of good contour or continuation X X X X O O O O (a) X X X X We perceive objects as continuous in cases where it is expected that they (b) continue 5. Principle of similarity X X O O X X X X O O X X O O X X O O O O X X O O X X O O X X X X O O X X (b) (a) and similarity (b)We organize objects or features of a scene based on similarity 6. Principle of common movement We group objects together that move together Principle of emergence Once “dog” is brought to mind, it becomes visible in this image (subconsciously following the principles as listed above) What do you see? Maybe it depends! ...
View Full Document

  • Winter '08
  • LEVITIN

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture