390 likes | 514 Views
Midterm 1. Oct. 6 in class Review Session after class on Monday. Read this article for Friday Oct 8th!. Mental Representations. Mental representations can start with sensory input and progress to more abstract forms
E N D
Midterm 1 Oct. 6 in class Review Session after class on Monday
Mental Representations • Mental representations can start with sensory input and progress to more abstract forms • Local features such as colors, line orientation, brightness, motion are represented at low levels How might a neuron “represent” the presence of this line?
Mental Representations • Mental representations can start with sensory input and progress to more abstract forms • Local features such as colors, line orientation, brightness, motion are represented at low levels • A “labeled line” • Activity on this unit “means” that a line is present • Does the line actually have to be present?
Mental Representations • Mental representations can be “embellished” - Kaniza Triangle is represented in a way that is quite different from the actual stimulus -the representation is embellished and extended
First Principles • What are some ways that information might be represented by neurons?
First Principles • What are some ways that information might be represented by neurons? • Magnitude might be represented by firing rate (e.g. brightness) • Presence or absence of a feature or piece of information might be represented by whether certain neurons are active or not – the “labeled line” (e.g. color, orientation, pitch) • Conjunctions of features might be represented by coordinated activity between two such labeled lines • Binding of component features might be represented by synchronization of units in a network
Visual Pathways • Themes to notice: • Contralateral nature of visual system • Information is organized: • According to spatial location • According to features and kinds of information
Visual Pathways • Image is focused on the retina • Fovea is the centre of visual field • highest acuity • Peripheral retina receives periphery of visual field • lower acuity • sensitive under low light
Visual Pathways • Retina has distinct layers
Visual Pathways • Retina has distinct layers • Photoreceptors • Rods and cones respond to different wavelengths
Visual Pathways • Retina has distinct layers • Amacrine and bipolar cells perform “early” processing • converging / diverging input from receptors • lateral inhibition leads to centre/surround receptive fields - first step in shaping “tuning properties” of higher-level neurons
Visual Pathways • Retina has distinct layers • signals converge onto ganglion cells which send action potentials to the Lateral Geniculate Nucleus (LGN) • two kinds of ganglion cells: Magnocellular and Parvocellular • visual information is already being shunted through functionally distinct pathways as it is sent by ganglion cells
Visual Pathways • visual hemifields project contralaterally • exception: bilateral representation of fovea! • Optic nerve splits at optic chiasm • about 90 % of fibers project to cortex via LGN • about 10 % project through superior colliculus and pulvinar • but that’s still a lot of fibers! Note: this will be important when we talk about visuospatial attention
Visual Pathways • Lateral Geniculate Nucleus maintains segregation: • of M and P cells (mango and parvo) • of left and right eyes P cells project to layers 3 - 6 M cells project to layers 1 and 2
Visual Pathways • Primary visual cortex receives input from LGN • also known as “striate” because it appears striped when labeled with some dyes • also known as V1 • also known as Brodmann Area 17
Visual Pathways • Primary cortex maintains distinct pathways – functional segregation • M and P pathways synapse in different layers W. W. Norton
The Role of “Extrastriate” Areas • Different visual cortex regions contain cells with different tuning properties
The Role of “Extrastriate” Areas • Consider two plausible models: • System is hierarchical: • each area performs some elaboration on the input it is given and then passes on that elaboration as input to the next “higher” area • System is analytic and parallel: • different areas elaborate on different features of the input
The Role of “Extrastriate” Areas • Functional imaging (PET) investigations of motion and colour selective visual cortical areas • Zeki et al. • Subtractive Logic • stimulus alternates between two scenes that differ only in the feature of interest (i.e. colour, motion, etc.)
The Role of “Extrastriate” Areas • Identifying colour sensitive regions Subtract Voxel intensities during these scans… …from voxel intensities during these scans …etc. Time ->
The Role of “Extrastriate” Areas • result • voxels are identified that are preferentially selective for colour • these tend to cluster in anterior/inferior occipital lobe
The Role of “Extrastriate” Areas • similar logic was used to find motion-selective areas Subtract Voxel intensities during these scans… …from voxel intensities during these scans …etc. STATIONARY STATIONARY MOVING MOVING Time ->
The Role of “Extrastriate” Areas • result • voxels are identified that are preferentially selective for motion • these tend to cluster in superior/dorsal occipital lobe near TemporoParietal Junction • Akin to Human V5
The Role of “Extrastriate” Areas • Thus PET studies doubly-dissociate colour and motion sensitive regions
The Role of “Extrastriate” Areas • Electrical response (EEG) to direction reversals of moving dots generated in (or near) V5 • This activity is absent when dots are isoluminant with background
The Role of “Extrastriate” Areas • V4 and V5 are doubly-dissociated in lesion literature:
The Role of “Extrastriate” Areas • V4 and V5 are doubly-dissociated in lesion literature: • achromatopsia (color blindness): • there are many forms of color blindness • cortical achromatopsia arises from lesions in the area of V4 • singly dissociable from motion perception deficit - patients with V4 lesions have other visual problems, but motion perception is substantially spared
The Role of “Extrastriate” Areas • V4 and V5 are doubly-dissociated in lesion literature: • akinetopsia (motion blindness): • bilateral lesions to area V5 (extremely rare) • severe impairment in judging direction and velocity of motion - especially with fast-moving stimuli • visual world appeared to progress in still frames • similar effects occur when M-cell layers in LGN are lesioned in monkeys
How does the visual system represent visual information? How does the visual system represent features of scenes? • Vision is analytical - the system breaks down the scene into distinct kinds of features and represents them in functionally segregated pathways • but… • the spike timing matters too!
Visual Neuron Responses • Unit recordings in LGN reveal a centre/surround receptive field • many arrangements exist, but the “classical” RF has an excitatory centre and an inhibitory surround • these receptive fields tend to be circular - they are not orientation specific How could the outputs of such cells be transformed into a cell with orientation specificity?
Visual Neuron Responses • LGN cells converge on “simple” cells in V1 imparting orientation (and location) specificity
Visual Neuron Responses • LGN cells converge on simple cells in V1 imparting orientation specificity • Thus we begin to see how a simple representation - the orientation of a line in the visual scene - can be maintained in the visual system • increase in spike rate of specific neurons indicates presence of a line with a specific orientation at a specific location on the retina • Why should this matter?
Visual Neuron Responses • Edges are important because they are the boundaries between objects and the background or objects and other objects
Visual Neuron Responses • This conceptualization of the visual system was “static” - it did not take into account the possibility that visual cells might change their response selectivity over time • Logic went like this: if the cell is firing, its preferred line/edge must be present and… • if the preferred line/edge is present, the cell must be firing • We will encounter examples in which these don’t apply! • Representing boundaries must be more complicated than simple edge detection!
Visual Neuron Responses • Boundaries between objects can be defined by color rather than brightness
Visual Neuron Responses • Boundaries between objects can be defined by texture
Visual Neuron Responses • Boundaries between objects can be defined by motion and depth cues