High-Level Vision

- Translational Invariance – can recognize object if it moves, not memorizing retinal image
- Size (Depth) Invariance
- Orientation (Shape) Invariance – rotation

- Description of object relative to the viewer
- If object moves, it has a new description
- If viewer moves, it has a new description

- Description of object relative to itself
- If object moves, description remains the same
- If viewer moves, description remains the same

people can more accurately identify flashed letter if incorporated into a word (top-down information)

- Difficulty recognizing variety of visually presented objects
- Normal recognition of objects through modalities other than vision
Intact visual sensation

- Inability to recognize faces
- Does visual face recognition work in different way than visual recognition of other types of objects?

if you’re good at recognizing something, holistic representation of object in mind; if we don’t have expertise, we analyze parts; humans are experts with faces

Patients seem virtually unaware of one region of space

individual features are extracted simultaneously (for the entire visual field at once) and automatically (without attention being focused on any one part of the visual field)

visual features are combined into representations of surfaces and objects (via the use of attention) – mental binding of features of object

- Individual visual features (e.g., motion) can be adapted – waterfall illusion (after watching waterfall, rock wall appears to be moving up, exhausting feature detectors)
- Segmentation “pop out”
- “Pop out” in visual search
- Pre-Attentive: search for single feature: parallel, level 1
- Attentive: search for conjunction of features: serial, level 2
- Illusory conjunctions: errors at level 2

- nature of pre-attentive processing (order seems to be opposite: Pre-attentive processing: late visual areas – obtain coarse gist of scene, Attentive processing: early visual areas – obtain fine details of scene, Requires feedback from late areas to early areas)
- Is search serial or parallel?

- Static image
- Processing occurs in parallel across image
- Progresses from gray-scale retinal input to internal representation in head

- Aim is to form a mental representation (no notion of an observer’s goal or task)
- Ignores inhomogeneity of retina
- Assumes trans-saccadic integration is flawless and easy
- Ignores eye movements in its theories of attention

- Emphasis on task – goal of vision is to enable observer to perform task – what you see is what you need
- Mental representation of visual scene is incomplete – representation emphasizes task-relevant details
- Blindness to task-irrelevant aspects of visual scene
- Relationship between vision and memory – Why form a mental representation of a visual object? If you need to know what an object looks like, then move your eyes and look at it

Genre - more wpm for light fiction than for biology
Age - more fixations per line for young children than for college students

modifications made to letters within 3-4 spaces to the left of fixation point or 15 spaces to the right of fixation point result in slower reading speeds

direction of asymmetry changes when reading Hebrew

- Observers position their eyes at each moment in time at the point in a scene that is currently most important for an on-going task
- Very few irrelevant areas are fixated
- “just-in-time” strategy – observers acquire specific visual info they need just at the point it is required in the task
- Eye movement patterns must be learned – expert/novice differences

Subjects view two images of objects, and must decide if the objects are the same or different. If the objects are the same, the images depict the object at different orientations. Reaction times are measured.

Roland and Friberg (1985) -- Increased blood flow in visual cortical areas during visual imagery task, but not during other cognitive tasks

Levine, Warach, Farah (1985) -- Patient with visual object identification difficulties was unable to draw familiar objects, despite being able to draw the relative locations of furniture in his house; Patient with visual object localization difficulties was unable when blindfolded to point to the furniture in his room, despite being able to give detailed descriptions of the appearance of familiar objects