Sensation and Perception Exam 2

Object and Scene Perception

The process of interpreting visual input into meaningful information that can be perceived

• Involves coding visual features into units and then interpreting those units as objects

Why is it difficult to emulate human abilities in machines?

Machines are unable to decipher every obstacle as being dangerous or not. They are not able to be coded for every detail.

• Ambiguity

• Hidden or blurred objects

Ambiguity

An object’s appearance changes because of different viewpoints, occlusion, and noise

Hidden/ Blurred Objects

Objects may be obstructed by other objects in the environment or can be blurred out.

Viewpoint Invarience

The ability to recognize an object regardless of viewpoint

Why Design a Perceiving Machine?

• Rescue robots

• Driving robots

• Surgical purposes

• Reduce accidents

Structuralism (Wundt)

Perceptions are created by combining elements called sensations (basic sensations)

Gestalt (Wertheimer)

Perception is a result of perceptual organization, perceiving the whole.

What two pieces of evidence is used to invalidate the structuralists theory of perception?

Apparent movement and illusory contours

Perceptual Organization

The process by which elements in a person’s visual field become perceptually grouped and segregated to create perception

2 Components of Perceptual Organization

• Grouping - putting together

• Segregating - see two separate things

Apparent Movement

An illusion of movement that occurs when two objects separated in space are presented rapidly, one after another, separated by a brief time interval (viewed as a whole)

Illusory Contours

Contour that is perceived even though it is not present in the physical stimulus (no physical edge)

• Neural data shows that illusory contours are processed in the V2 (secondary visual cortex)

Von der Heydt

• Showed that edge detection cells respond to the illusory edges as strongly as real edges

• But did not fire if no edge was implied

Good Continuation

Connected points resulting in straight or smooth curves that are seen as belonging together

Principle of Prägnanz

Perceive complex things as simplified forms to easily recognize and understand what they say, because it is the interpretation that requires the least cognitive effort from us (Prägnanz = “good figure” or “pithiness”)

Similarity

Similar things appear to be grouped together

Proximity

Things that are near to each other are grouped together

Common Fate

Objects move in the same direction as a group

Figure Ground Segregation

Determining what part of the environment is the figure so that it “stands out” from the background

• Figure: More “thinglike” and memorable than the ground; seen in front of the ground

• Ground: More uniform and extends behind the figure; surrounds the figure

Findings from Vercera et al. (2002)

• Information within the image determines perception grouping

• Areas lower in the field of view are more likely to be perceived as a figure

  • A figural cue for segregation

Findings from Max Wetheimer (1912)

• A - seen as a W & M, using our knowledge to perceive

• B - seen as a pattern, so principles of continuation override knowledge (connected points)

• Built-in perceptual organization can override knowledge

Findings from Gibson & Peterson (1994)

• Figure-ground formation can be affected by the meaningfulness of a stimulus

• The black part is seen as a figure when upright and appears like a woman, but does not appear that way when it’s viewed upside down

• Meaningfulness can influence identifying a figure in an image

Recognition by Components

• A theory stating that object recognition occurs by representing each object as a combination of basic units

Geons

• The visual system breaks down objects into geometric units or geons

• Geons (40) - basic units of objects consisting of simple shapes such as cylinders and pyramids

Recognition by Components - Advantages and Disadvantages

• Advantage

  • Accounts for viewpoint invariance - objects are seen as the same regardless of the vantage point

• Disadvantages:

  • Doesn’t account for all types of object recognition, like face recognition, oddly shaped objects (like clouds)

  • Doesn’t allow for fine-grained discriminations (two types of dogs have the same shape but are different breeds)

Scene Perception

• A scene contains background elements and objects organized in meaningful ways with each other and the background

• Scene - acted within

• Objects - acted upon

Potter (1976) - Method

• Showed that people can perceive the gist of an image when it is presented for 250 ms

• She first presented either a target photograph or, as shown here, a description, and then rapidly presented 16 pictures for 250 ms each. The observer’s task was to indicate whether the target picture had been presented

Masking Procedure

A stimulus is covered by a random pattern to eliminate the persistence of vision

• Persistence of vision: perception of any stimulus persists for about 250 ms after the stimulus is physically terminated

Fei-Fei (2007)

• Used masking to show that the overall gist is perceived first, and then followed by details

  • 27 ms - tell between dark and light

  • 67 ms - identify large objects

  • 500 ms - smaller objects and details

Fusiform Face Area (FFA)

A region in the brain’s interior temporal lobe that is responsible for processing faces in the fusiform gyrus

Nancy Kanwisher et al. (1997) - Functional Localizer

• Identifying the region of interest (ROI) with fMRI

• “The only region in which most of our subjects (12/15) showed a significantly greater activation for faces than objects was in the right fusiform gyrus.”

Confounds in Nancy’s Study

Luminance, Categories, Human Body Parts, and Orientation of the Face

What type of representation does Nancy’s study support?

The idea of a specialized area only meant for one thing, supports the hypothesis of modularity

Parahippocampal Place Area (PPA)

Located in the temporal lobe, the PPA processes visual scenes and environments (landscapes, buildings, rooms)

Tong et al. (1998)

The procedure used for binocular rivalry is to present a photo of a person's face to one eye and a photo of another object( such as a house or a bike) to the other eye. Tong et al. used colored glasses to allow the face to present only in one eye and the object in the other. The results of this procedure displayed that changes in perception and changes in brain activity mirrored each other. The two important brain areas highlighted in this research are the fusiform face area (FFA) and parahippocampal place area (PPA) because when faces were perceived the FFA had visible activity and when the houses were shown the PPA would show activity. This finding is important because it provides evidence as to how the brain consciously perceives faces versus objects or places. 

Binocular Rivalry

The observer perceives either the left or the right image, but not both at the same time

Expertise Explanation of the FFA Area

• Novice and Expert groups

• Brain scans taken while looking at faces and greebles

  • Greebles - artificial objects designed to be used as stimuli in psychological studies of object and face recognition

• Novices show less greeble FFA activity than the experts

• Isabel Gauthier showed we are experts of faces

Voxel

A little cube of brain area, and each voxel contains a single number representing the signal measured at that location

Haxby et al. (2001) - Within vs Between

• Wanted to see whether the pattern response in the ventral vision pathway could be distinguished from the patterns of responses by all of the other categories

• Within Category Correlation

  • to determine the pattern of responses to a specific exemplar

• Between Category Correlation

  • to determine the pattern of responses to one category

• If the within correlations are bigger than the between correlations, then that was used as evidence for distinct patterns of neural responses, which was true for all categories

Haxby et al. (2001) - Correlations

• Got rid of the most active voxels

• Within correlations were not reduced that much even with the FFA voxels removed

• Holds true for the other categories as well

Haxby et al. (2001) - Results

Distributed representation of categories - Haxby et al. (2001) provided evidence of a distributed representation for faces, because even with the FFA removed there was still a large within correlation

Attention

Process of focusing on a specific object while ignoring others

Overt vs Covert Attention

• Overt: looking directly at the attended object

• Covert: attention without looking

Selective Attention

Focus on a specific object or activity while ignoring distractions (can be overt or covert)

Spatial Attention

Attended to a specific location in space (can be overt or covert)

Dichotic Listening Task - Cherry (1953)

The participant listens to different audio inputs in different ears and shadows back what they are hearing

Shadowing

Repeat what you are hearing

Finding from the Dichotic Listening Task

• Also known as the cocktail party effect

• The attended ear allows the participant to shadow that they hear, but the unattended ear will not remember specific details but will remember physical characteristics of the audio input.

ADHD and Dichotic Listening

• ADHD group was impaired at reporting words in the selected ear as compared to controls

• Demonstrating a deficit in selective attention

Broadbent Filter Model of Attention

• Cherry’s work led to this early attention theory

• How it is possible to focus one message

Posner (1978) - Pre-Cueing Procedure

The precueing procedure starts with the participants looking at a fixation point to prevent overt attention and then precue with an arrow. Stimuli is either consistent (valid) or inconsistant (invalid). Valid trails are when you are pre-cued to the side of the target which in turn produce a faster reaction time. Invalid trials are when the cue indicates a target will appear in a location that is different from where it actually does.

Posner (1978) - Findings

Information processing is more effective at the place where attention is directed because it enhances information at a location. Invalid cues lead to a slower response because the brain needs to suppress the incorrect response and initiate the response for the right side.

What does Posner’s Study Demonstrate about Spatial Attention?

• Helps us to understand attentional processes like alerting (readiness to respond) or orienting (selection)

• Important for navigating our world

Spatial Attention and Aging

• “Slower responses in the invalid relative to the valid condition became greater with the increase in age.”

• It takes older adults longer to process the invalid cue and reorient their attention

Binding

Process by which features are combined to create a perception of coherent objects

Binding Problem

Features of objects are processed separately in different areas of the brain

Illusory Conjunctions

Features that should be associated with an object become incorrectly assocaited with another

Triesman and Schmidt (1982) - Findings

• Found the effect of divided attention on feature integration

• Divided attention: completing multiple tasks at once, like having to report the numbers and shapes

• Incorrect associations occurred 18% of the time

Feature Integration Theory

Explains how an object is broken down into features and how these features are recombined to result in perception

2 Stages of Feature Integration Theory 

• Pre-attentive stage: features of objects are separated

• Focused attention stage: features are bound into a coherent perception

Visual Search

Looking for an object among the objects

Feature Search (parallel) vs Conjunction Search (serial)

• Feature Search (parallel): when the target can be found by one feature (green line)

  • Rapid; pop-out effect; does not require attention; everything is processed simultaneously (parallel)

• Conjunction Search (serial): search for two features (horizontal and green line)

  • Slower; needs attention; search serially

Balint Search

• Balint’s Syndrome: parietal lobe damage, which reduces the ability to shift and focus attention

  • A patient could complete task A, but failed task B

  • Couldn’t combine features with attention

  • Attention at a particular location isn’t required for single feature detection

  • Proves illusory conjunctions

Visual Scanning - Fixations vs Saccades

• Each time you paused on one face, you were making a fixation, which allows us to focus

• When you move your eyes to another face, then you make a saccadic eye movement - a rapid eye jerky movement from one fixation to the next

What directs our attention?

Stimulus salience: physical properties of an object that direct our attention to that object (e.g., color, loudness, contrast, movement)

Attentional Capture

Process of involuntarily directing our attention to a particular object because of stimulus salience

Saliency Map

Characterizes the saliency of objects or areas within a scene

Parkhurst et al. (2002)

• Initial fixations are determined by saliency

• Subsequent fixations are determined by top-down factors, such as observer interest

Selection Based on Cognitive Factors

• Picture meaning and observer knowledge

  • Fixations are influenced by this meaning or knowledge

• Task demands: can affect the directing of attention

• Yarbus (1967): eye movements change depending on the goal

Inattentional Blindness

The failure to see fully visible objects or events in the visual display because one’s attention is focused elsewhere

• Example: Gorilla video (46% fail to notice visually salient gorilla)

Change Blindness

The failure to detect obvious changes in a scene when vision has been disrupted

• Flicker test with masking

J.J. Gibson

• Questioned the traditional lab approach

• Felt it was too artificial (observers weren’t allowed to move their heads)

• Unable to provide an explanation for many things (e.g., how pilots use their environment to land planes)

• Suggested we needed to investigate senses together, not individually (balanced)

• Came up with the ecological approach

Ecological Approach

• Emphasizing the study of moving observers to determine how their movement results in perceptual information that both creates perception and guides further movement

Ecological Validity

When an experiment’s stimuli, conditions, and procedures match the natural world

Optic Flow

The apparent motion of objects as the observer moves past them (perceptual cue processed for self-motion control)

Gradient Flow

Provides information about how fast we are moving

• Faster near observer and slower further away (provides clues about our speed)

• Change in this relative difference provides information on speed (smaller difference between far away and near equals a slow speed)

Focus of Expansion

No flow at the destination point

• This is a type of invariant information: remains constant regardless of whether the person changes their final destination point. It will always be perceived as a non-moving point

Self-Produced Information

When a person makes a movement, that movement provides information, which guides further movement (reciprocal relationship)

Bardy and Laurent (1998)

• Study on novice and professional gymnasts doing somersaults with their eyes closed

• The vestibular system is the part of the inner ear that controls balance

• Maintaining balance is difficult with your eyes closed, because vision works with the vestibular system, which provides information to help your muscles make adjustments

Lee and Aronson (1974)

• Swinging room experiment

• Demonstrated how visual information influences balance by overriding the vestibular system because no actual movement was occurring

Walking - Visual Direction Strategy

Observers keep their bodies pointed toward a target (correcting themselves when they drift left or right)

Blind Walking

• Shows that people can navigate without any visual stimulation from the environment

• Philbeck (1997): “Blind walking” procedure

  • People observe a target object located up to 12 meters away, then walk to the target with their eyes closed

  • Accomplished by combining memory of position with knowledge of movements

• Observers use spatial updating, which is keeping track of their position as they move

Way Finding

Navigating through an environment (like driving from Austin to LA without a GPS)

Landmarks

Objects on the route that serve as cues to indicate where to turn

Hamid et al. (2010) - Findings

• Removing landmarks hinders wayfinding

• Non-decision vs decision landmarks

• Landmarks looked at most to help with navigation, when removed, then there was a drop in performance

Jazen & van Turennout (2004) - Findings

• Greatest brain activation for objects at decision points (landmarks) was in the parahippocampal gyrus (spatial memory and navigation) at test

• Brain encodes landmarks at decision points automatically, even if they aren’t remembered

Brain GPS

• Tolman (1930s and 1940s)

• Rats created a cognitive map when exploring a maze

  • Cognitive map: a mental map of the spatial layout of an environment

• They did not just learn turn right for food

Place Cells

• O’Keefe (1970s): discovered place cells in the hippocampus

• Place cell: fires when an animal is in a certain place in the environment

  • Only in the physical place, not just looking at it

Grid Cells

• Moser & Moser (2000s): discovered grid cells in the entorhinal cortex

• Grid cell: fires when an animal is in a certain place in the environment and has multiple place fields arranged in regular, grid-like patterns (hexagon) - creates a cognitive map

  • Fires at regular intervals as an animal navigates an open area

  • Allowing the animal to understand its position in space by storing and integrating information about location, distance, and direction

• Forms a unique pattern of coordinates, which is shifted with respect to the coordinates formed by other nearby grid cells

• The whole environment is “filled” with grid patterns

• Each grid cell is active in multiple locations

Parietal Reach Region (PRR)

In the parietal cortex, it is involved in reaching for objects (contains neurons for control of reaching and grasping)

Mirror Neurons

A class of neurons that modulate their activity both when an individual executes a specific motor action and when they observe the same or similar action performed by another individual

• Responds when an animal grasps an object and when viewing someone grasp an object

• PPR responds to the observed action, “mirrors” the response of actually grasping

• Has a diminished response if grasped by a tool, indicating it is specialized to the type of motion, not the pattern

Audio Visual Mirror Neurons

Respond to action and the accompanying sound (in premotor cortex)

• Kohler et al. (2002):

  • Hearing or watching a peanut being broken caused brain activity that is associated with the action

  • Audiovisual neurons respond to “what is happening”, not just the pattern of movement

Mirror Neurons and Intentions

• Iacoboni et al. (2005)

• Hypothesis: If mirror neurons are only influenced by actions, then the control action and intention action should have the same response in the brain

• Found more activation in the mirror neuron networks when there was intention

  • Mirror neurons can be influenced by different intentions

  • Mirror neurons encode the “why”

Possible Functions of Mirror Neurons

• To help understand another animal’s actions (intentions) and react to them appropriately

• To help imitate the observed action

• May help link sensory perceptions with motor actions

• Implicated in communication, empathy, etc.