Ashcraft, H. M. & Radvansky, A. G.: Cognition (5. udg.). Pearson International Edition, Prentice Hall. 2010 (svarende til 335 normalsider).
Light waves enter the eye, are focused and inverted by the lens, and are projected onto the retina.
The Retina is composed of three layers of neurons: Rods and Cones, Bipolar Cells and Ganglion Cells.
The rods and cones form the back layer of neurons and are stimulated by light. They begin the process of vision → neural firing from the rods and cones are passed to the second layer, the bipolar cells, which collect the messages and move them along to the third layer: the ganglion cells. The axons of the ganglion cells converge at the rear of the eye, forming the bundle of fibres that is the optic nerve.
Each eye transmits to the occipital lobes of both hemispheres.
Each half of the retina gathers information from the contralateral visual field.
Fixation Point: Where you are looking.
Compression: A transformation that analyses and summarizes visual input.
The message that finally reaches the visual cortex represents an already processed and summarized record of the original stimulus.
There are approximately 120 million rods on each retina and about 7 million cones.
Most of the cones lie in the small area known as the fovea, which provides us with out most accurate, precise vision.
In the fovea one cone connects with one bipolar cell (a cone synapses onto a bipolar cell), but in peripheral vision, tens or even hundreds of rods converge on a single bipolar cell.
About 1 million ganglion cells combine to form the optic nerve.
Sensation: The reception of stimulation from the environment and the initial encoding of that stimulation into the nervous system.
Perception: The process of interpreting and understanding sensory information.
Gathering Visual Information
– Vision is triggered when the reflection of light from an object hits our eyes.
– The eye sweeps from one point to another in fast movements called saccades, movements that are interrupted by pauses called fixations.
– For the most part we only take in information during a fixation.
– Visual attention should be interruptible but not too interruptible.
Change Blindness: Failure to notice changes in visual stimuli, when those changes occur during a saccade.
Inattentional Blindness: Failure to see an object we are looking at directly, even a highly visible one, because our attention is directed elsewhere
Visual Sensory Memory
Visual Sensory Memory (Iconic Memory): A temporary visual buffer that holds visual information for brief periods of time.
Visual Persistence: The apparent persistence of a visual stimulus beyond its physical duration.
Amount and Duration of Storage
Sperling et al:
ñ Span of Apprehension (The Span of Attention/The Span of Immediate Memory): The number of individual items recallable after any short display.
ñ Whole Report Condition: People are to report any letters they can. The whole display was to be reported.
ñ Partial Report Condition: Only one of the rows was to be reported.
Icon: An image in iconic memory – the visual image that resides in iconic memory.
Erasure and Interference
Decay: Forgetting as a passive process like fading. The mere passage of time degraded the icon, making it illegible after a short interval.
Interference: Forgetting caused by the effects of intervening stimulation or mental processing.
Backward Masking: A later visual stimulus can drastically affect the perception of an earlier one.
Erasure: When the contents of visual sensory memory are degraded by subsequent visual stimuli, the loss of the original information is called erasure, a specific kind of interference.
Beta Movement: The brain fills in any jumps in position, producing the illusion of motion. A mental perceptual inference of illusory motion. Occurs when making inferences from one picture to the next.
Phi Phenomenon: When iconic memory receives visual images in relatively close proximity in space and time, it will infer a virtual movement. Involves illusory tracking of an object in space.
The Early Parts of a Fixation
Iconic memory is the first phase in visual perception.
Dynamic Icons: Iconic images that contain movement.
Temporal Integration: Perceiving two separate events as if they had occurred at the same time – happens when visual events occur within about 20 ms of each other.
A Summary for Visual Sensory Memory
The duration of a normal iconic memory, persistence: 250-300 ms.
Focal Attention: Mental process of visual attention, such as the mental redirection of attention when the partial report cue is presented.
The memory system that is used across a series of eye movements. Works using object files. To build up a complete mental representation of the world we use trans-saccadic memory.
Retinal coordinates: Where the images fall on the eyes.
Spatial coordinates: Where things are in space.
Object files: Representations of individual objects that iconic memory uses to track what is going on in the world.
The role of visual sensory memory is to encode the visual information into the memory system, so that pattern recognition can take place.
Gestalt Grouping Principles
Principles of perceptual organization laid out by the Gestalt psychologists in the early to mid 20th century. They identify those characteristics of perception in which ambiguities in a stimulus are resolved to determine which entities are present.
Figure-ground principle: When viewing an image, part of the image will be treated as the figure or foreground (the object identified), which is segregated from the visual information upon which this object is set (the background).
The Gestalt Grouping principles are aimed at providing a more complete percept from incoming image information that may be fragmentary or incomplete.
Closure: A person “closes up” an image that has gaps or parts missing.
Proximity: Elements that are near one another tend to be grouped together in perception.
Similarity: Elements that are visually similar in some way, such as having similar colour or texture, tend to be grouped together.
Good Continuation: When there is an edge that is occluded or interrupted, people will assume that it continues along in a regular fashion.
Common Fate: Entities that move together are perceptually grouped together.
The Template Approach
Templates: Stored models of all categorizable patterns.
We have preferred viewing angles for many objects, the canonical view.
Visual Feature Detection
Feature Detection (Feature Analysis): A feature is a very simple pattern, a fragment or component that can appear in combination with other features across a wide variety of stimulus patterns.
Feature theories claim that we recognize patterns by first identifying the building-block features.
ñ Pandemonium, Selfridge 1959.
Reigns in the process of pattern recognition because of the mental mechanisms that process a visual stimulus. These mechanisms are demons. Pattern is encoded by a set of data demons. Next the computational demons begins to act. These are the feature analyzers, and when they match a stimulus feature, they begin to shout excitedly. The cognitive demons listen to all this shouting. The cognitive demons represent the different letters of the alphabet, one for each letter. Any evidence from the computational demons that suggests a match with the stimulus causes the cognitive demon to begin shouting as well. The one who shouts the loudest is the one whose pattern is most nearly matched by the input stimulus. The loudest cognitive demon is finally heard by the decision demon, the highest-level demon in the model. The demon has the final say in recognizing and categorizing the pattern.
→ Neurophysiological studies showing that specialized visual cortex cells exist for various simple visual features and patterns.
→ Feature detection may have a physiological status in the nervous system. (Hubel and Wiesel)
→ Feature detection is probably a simultaneous or parallel process instead of a serial process.
→ Perception is a problem solving process.
Beyond Features: Conceptually Driven Pattern Recognition
Pandemonium was a completely bottom-up processing system – a completely data-driven processing system where processing is driven by the stimulus pattern, the incoming data. It is missing context: A mechanism that would allow context and a person’s expectations to influence the recognition of patterns. Such effects are called top-down processing or conceptually driven processing effects, where context and higher-level knowledge influence lower-level processes.
Repetition Blindness: The tendency to not perceive a pattern, whether a word, a picture, or any other visual stimulus, when it is quickly repeated. (Morris and Harris)
Misreading Effect: A tendency to read a word that should be in the sentence, based on the context, rather than the one that actually appears.
Connectionist Modeling, p. 90-91.
A theoretical and computational approach to some of the most challenging issues in cognitive science. Connectionist models involve a massive number of mathematical computations.
Usually three levels of units:
ñ The Input Level
ñ The Hidden Level (invisible to an outsider)
ñ The Output Level
These levels are interconnected, and the connections are either positively or negatively weighed.
Excitatory and inhibitory neurotransmitters.
Neural Net Modeling
Input Units: Extremely basic, elementary “cells” in the structure, which receive the inputs from the environment.
When a stimuli is presented to the input device, one or more of the input units matches the features in the stimulus. Then each unit that matches activates a set of connected units in the middle level of the structure, the hidden unit level: hidden here simply means that this level is completely internal, always one step removed from either input or output.
The connections always have a weight attached to them, a weight that represents the relationship between the linked units. Hidden units that receive enough positive activation, called excitation, govern the outcome of processing. Units receiving negative activation, or inhibition, end up having little control over the outcome. After all the weights have been factored into the computational formulae, activations at the output level come into play.
Output Units: The units that report the system’s response.
Higher-level knowledge is participating in the lower-level task of identifying letters.
Object Recognition and Agnosia
Recognition by Components
Biederman: We recognize objects by breaking them down into their components, then looking up this combination of components in memory to see which object matches the combination. The human recognition system has a small number of basic primitives, simple three-dimensional geometric forms called geons, a combined form of geometric ions.
Recoverable and non-recoverable drawings. Depends on which lines are degraded. If segments of the smooth, continuous edges are missing, it is relatively easy to fill in the missing parts from memory and to recognize the pattern. If the parts that are missing are in important locations, where the vertices are missing, recognition is much more difficult.
Shortcomings of Recognition by Components and Embodied Perception
The model is tied to bottom-up processing, but there is now ample evidence that object recognition is strongly influenced by context and prior knowledge. There are also data that show that people can perceive the overall shape and pattern of an object as rapidly and accurately as they perceive the components.
Our perception of objects can be influenced by our expectations of how we will interact with those objects: Embodied cognition can influence perception. Emotions can also meaningfully influence perception.
A failure or deficit in recognizing objects, either because the pattern of features cannot be synthesized into a whole or because the person cannot then connect the whole pattern to meaning. People can no longer perform the seemingly instantaneous mental steps of pattern recognition.
It is a cognitive mental loss: the agnosic can input the basic visual stimulus but cannot do anything with that encoded information.
Prosopagnosia: A disruption of face recognition. When the disruption affects people’s recognition of faces, sometimes while leaving object recognition intact.
Dr. P: The most famous case of agnosia. Dr. P had lost his ability to recognize objects and faces.
Apperceptive Agnosia: A basic disruption in perceiving patterns. The ability to coalesce basic visual information into a percept, an entity, or a whole is lost.
Associative Agnosia: The person is able to construct a mental percept, but he or she cannot associate the pattern with meaning.
Implications for Cognitive Science
There is growing evidence that the right hemisphere is more involved in global processing, to include forming global patterns, and that the left hemisphere plays more of a role in local processing (processing small components and features).
The temporal lobes are particularly associated with areas related to language and word meaning.
- Detecting the features in a visual stimulus is a separate (and later) process from the sensory steps that encode a stimulus into cognition.
- Detecting the visual features is critical in constructing a perceived pattern, a percept. If the stimulus features cannot be extracted, then the person cannot form an overall pattern or concept.
- There is a separate step involved in hooking up the pattern with its meaning and name, involving the visual association from the pattern to the knowledge stored about it in memory.
Auditory stimuli consist of sound waves moving in the air. The human auditory mechanism that responds to these stimuli translates the sound waves into a neural message. From the elements of funnels, moving bones and fluid arises our sense of hearing or audition.
Auditory input to the brain is sent primarily to the auditory cortex. The human hearing is almost unbelievably complex. We can discriminate accurately between highly similar sounds even from birth, and we routinely convert the continuous stream of sounds known as speech into a comprehended message with little or no apparent effort.
Auditory Sensory Memory and Echoic Memory are used interchangeably. They both refer to a brief memory system that receives auditory stimuli and preserves them for some amount of time.
The function of echoic memory is to encode sensory stimulation into the memory system and hold it just long enough for the rest of the mental system to gain access to it.
The Modality Effect: Superior recall of the end of the list when the auditory mode is used instead of the visual mode of presentation.
The Suffix Effect: Inferior recall of the end of the list in the presence of an additional, meaningful, non-list auditory stimulus. The more the suffix is like the information on the list, the greater the suffix effect.
The auditory sensory system register sensory information and hold it for 2,000 to 4,000 ms.
Problem of Invariance: The sounds of speech are not invariant from one time to the next.
Context plays an important role in spoken word identification.
Phoneme: Language sound
Perception and identification of speech are heavily dependent on context, on top-down processing.