ch.1.docx

16 Pages
95 Views

Department
Psychology
Course Code
PSYB57H3
Professor
Michael Inzlicht

This preview shows pages 1,2,3,4. Sign up to view the full 16 pages of the document.
Description
Chapter 1 THE SCOPE OF COGNITIVE PSYCHOLOGY o actions, thoughts, and feelings depend on knowledge o Many/all encounters with world depend on supplementing experience with knowledge that you bring to the situation  H.M; his memory loss unanticipated product of brain surgery intended to control epilepsy, lived > 50 years after  could remember events prior to surgery, unable to recall event that occurred after his operation  Without a memory there is no self o Self-concept depends on knowledge (episodic knowledge); our emotional adjustments to the world, rely on our memories A BRIEF HISTORY The Years of Introspection th o 19 C, Wundt and Titchener new enterprise of research psychology, separate from philosophy or biology o Study thoughts need us to introspect, “look within” to observe/record content of our own mental lives, sequence of experiences o Introspectors trained: given vocabulary to describe what they observed; report on experiences, with no interpretation o Introspection, study of conscious experiences, tell us nothing about unconscious events o If we rely on introspection as means of studying mental events, we have no way of examining these processes o Problem with introspection: in order for any science to proceed, must be way to test its claims; otherwise no means of separating correct assertions from false ones, accurate from fictions - unattainable in introspection o Science needs objective observations that aren’t dependent on a particular point of view or a particular descriptive style Behaviourism o Psychology needs objective data, observeable o Behaviour = objective data, scientific mill; stimuli in world are in same objective category: measurable, recordable, physical events o Learning history can be objectively recorded and scientifically studied- beliefs, wishes, goals, expectations are things that cannot be directly observed/objectively recorded; need to rule out any discussion of “mentalistic” notions; observed only via introspection o Behaviourist movement dominated psychology in America, 1 half of the 20 C: uncovered range of principles concerned with how behaviour changes in response to various stimuli o Principles remain in contemporary psychology, basis for learning theory, as well as a wide range of practical applications o Ways people act, feel, are guided by how they understand or interpret the situation, and not by the objective situation itself o Behaviourist’s perspective demands that we not talk about mental entities such as beliefs, memories, etc. B/c these things cannot be studied directly and therefore cannot be studied scientifically Cognitive Revolution o Nearing an impasse: if we wish to explain/predict behaviour, we need to make reference to mental world of perceptions, understanding, and intentions; how people act is shaped by how they perceive the situation, how they understand the stimuli o A solution to this impasse, suggested by Immanuel Kant – transcendental method, begin with observable facts, then work backward from these observations; method sometimes called “inference top best explanations” is at the heart of most modern science RESEARCH IN COGNITIVE PSYCHOLOGY: AN EXAMPLE Working Memory: Some Initial Observations o Working memory: information you are actively working on; holds info in easily accessible from “at your fingerprints”, instantly available o Hypothesized to have a small capacity, only a few items held in this store, you will never have a problem locating just the item you want o Span test; we read to someone a list (letters); person has to report these back, immediately in sequence; if person succeeds, we try it again with 5 letters until you find a list that person cannot report back; people start making errors with sequences of 7/ 8 Working Memory: A Proposal o When people make mistakes in this task, they substitute one letter for another with a similar sound; having heard “S” report back “F” o Problem is not hearing the letters in the first place: we get similar sound-alike confusions if the letters are presented visually; having seen “F” people are likely to report back “S” o Several different parts, working-memory system; heart of system is central executive, does the real work o Working memory’s assistants provide the same benefit and one of the most important assistants is the articulatory rehearsal loop- trying to remember 5 #; # maintained by working memory’s rehearsal loop, with # out of way, central executive was free to continue reading; advantage of this system” with storage handled by helpers, the executive is available for other, more demanding tasks o to launch the rehearsal loop, rely on process of subvocalization, or silent speech; i.e. saying the numbers quietly to yourself o subvocalization produces a representation of these # in phonological buffer; an auditory image is created in inner ear o rehearsal loop requires two elements, working tandem: subvocal speech (inner voice) and the phonological buffer (inner ear) o the executive isn’t needed to maintain these materials and is free to focus on other activities Evidence for the Working Memory System o memory items are briefly stored as (internal representations of) sounds, no surprise that errors are shaped by this mode of storage o Concurrent articulation task requires mechanisms for speech production; these mechanisms are not available for other use, including subvocalization. If your directing your lips and tongue to produce “Tah-Tah-Tah” sequence, you can’t at same time direct them to produce the sequences needed for the subvocalized materials o Our original span test measured combined capacities of the central executive and the loop; when people take a span test, they store some of the to-be-remembered items in the loop and others via the central executive o With concurrent articulation, loop isn’t available for use and now measuring capacity of working memory without rehearsal loop o Concurrent articulation, even though it’s extremely easy should cut memory span drastically; span is ordinarily about 7x; with concurrent articulation it drops roughly a third – 4/5x o With visually presented items, concurrent articulation should eliminate sound-alike errors; repeatedly saying TAH-TAH blocks the use of the articulatory loop, and it is in this loop that the sound-alike errors arise o With concurrent articulation and visual presentation of the items, sound alike errors are largely eliminated o If people aren’t using rehearsal loop, there should be no cost attached to denying them use of the loop o Rehearsal loop required only for storage; like all of the working memory assistants is incapable of any more sophisticated operations o Concurrent articulation blocks use of loop but no effect on someone’s ability to read brief sentences, to do simple logic problems etc. o blocking use of loop does have an effect when you’re reading more complex sentences or doing harder problems b/c these harder tasks require analysis and the storage of interim steps and so require the entire working-memory system – the executive and the assistants The Nature of the Working-Memory Evidence o cognitive neuroscience – study of the biological basis for cognitive functioning o b/c of specific forms of neurological damage, some individuals have no ability to move these various muscles - suffer from anarthria – an inability to produce overt speech o data indicate that these individuals, show sound-alike confusions in their span data, just as ordinary participants do; word-length effect associated with the use of the rehearsal loop  actual muscle movements aren’t needed to subvocalization, b/c the results are the same without these movements o inner speech relies on brain areas responsible for planning/controlling muscle movements of speech, not on movements themselves o observations from neuropsychology, concerned with how various forms of brain dysfunction influence observed performance o deaf rely on different assistant for working memory: they use an inner hand and covert sign language rather than an inner voice which is a covert speech; they are disrupted if they are asked to wiggle their dingers during a memory task and they also tend to make same hand shape errors in working memory (analogues to the sound-alike errors made by the hearing population) Working Memory in a Broader Context o you rely on working memory in a vast number of circumstances; and if we understand working memory, we move toward an understanding of this far broader set of problems and issues o without the option of relying on articulatory rehearsal, reading becomes much more difficult when someone is learning new vocabulary, including vocabulary in a new language o working memory and articulatory rehearsal are relevant to a wide range of mental activities in adults and in children o understanding working memory may give us insight into a broad range of tasks Chapter 2 Capgrass Syndrome: An Initial Example  Rare; seems to be one of the accompaniments to Alzheimer’s disease and is sometimes observed in the elderly  The people with this are convinced that the people they know are not who they appear to be  They think that their family members have been replaced by imposters  This can lead to paranoid suspicions about why a loved one has been replaced and why no one is willing to acknowledge this replacement  Facial recognition involves two separate systems o Cognitive appraisal o Emotional appraisal  The emotional appraisal is disrupted in this syndrome leading to the intellectual identification without the familiarity response The Neural Basis for Capgras Syndrome  Current studies rely on MRI scans  Right temporal lobe is damaged in Capgras patients and this damage disrupts circuits involving the amygdala  Amygdala serves as an emotional evaluator helping an organism to detect stimuli associated with threat or danger and for detecting positive stimuli  Capgras patients also have damage in the prefrontal cortex  fMRI allow us to track moment-by-moment activity levels in different sites in a living brain  prefrontal cortex is active when person is engaged in tasks that require planning, or careful analysis and it is less active when dreaming  Capgras patients are less able to keep track of what is real and what is sensible; weird beliefs can emerge unchecked, including delusions What Do We Learn From Capgras Syndrome?  Damage to the amygdala is probably the reason Capgras patients experience no sense of familiarity when they look at faces they know well  The damage to prefrontal cortex helps us understand why Capgras patients offer crazy hypotheses about their skewed perception  Capgras syndrome can be used to illuminate broader issues about the nature of the brain and of the mind  It suggests that the amygdala plays a crucial role in supporting the feeling of familiarity and the biological evidence suggests that the amygdala also plays a central part in helping people remember the emotional events of their lives The Study of the Brain  Brain weighs about 3-4 pounds  Contains large number of glial cells  Phineas Gage damage to front-most part of his brain led to severe personality and emotional problems Hindbrain, Midbrain, Forebrain  Hindbrain has structures crucial for controlling key life functions and also plays an essential role in maintaining the body’s overall tone o It helps maintain the body’s posture and balance and helps control the brain’s level of alertness o Largest area of hindbrain is cerebellum and damage to this can cause problems in spatial reasoning, in discriminating sounds, and in integrating the input received from various sensory systems  Midbrain plays important role in coordinating movement and the circuits that relay auditory info from the ears to the areas in the forebrain where this info is processed and interpreted  The largest region is the forebrain o The cortex is the outer surface of the forebrain o Cortex constitutes 80% of the human brain o The convolutions are the brain’s most obvious visual feature o Longitudinal fissure = deepest groove which separates the left cerebral hemisphere from the right o Central fissure divides the frontal lobes on each side of the brain from the parietal lobes o The bottom edge of the frontal lobes is marked by the lateral fissure, and below it are the temporal lobes o At the very back is the occipital lobe Subcortical Structures  Underneath the cortex are the subcortical parts of the forebrain  Thalamus = relay station for all sensory info going to cortex  Hypothalamus = plays crucial role in controlling motivated behavior  Limbic system = include the amygdala and hippocampus; essential for learning and memory Lateralization  Cortical and subcortical structures appear in both sides of the brain in roughly the same structure  Commisures connect the 2 hemispheres and carry info back and forth from one hemisphere to the other o Largest commissure is the corpus callosum Data from Neuropsychology  Neuropsychology = study of the brain’s structures and how they relate to brain function  Symptoms developing from brain damage depend heavily on the site of the damage  A lesion in the hippocampus produces memory problems but not language disorders  Damage to left side of frontal lobe = disruption of language use  Different brain areas produce different functions Data from Neuroimaging  CT scans study brain structure while PET scans study brain activity  CT relies on X-ray and provides 3D image  PET uses a tracer substance and tells which tissues are using more glucose  CT scans tell size, shape and position of structures within brain and PET tells which regions are active at any point in time  MRI uses magnetic properties of atoms that make up brain tissue  fMRI measures the oxygen content in blood flowing through each region of the brain (precise pictures)  CT and MRI scans are stable while PET and fMRI change with activity Data from Electrical Recording  Because millions of neurons are active at the same time, the current generated by them is enough to be detected by electrodes on the scalp (EEG)  EEGs are used to study broad rhythms in the brain’s activities  To measure the changes in EEG in a brief period just before, during and after an event are changes referred to as event-related potentials (ERPs) The Power of Combining Techniques  Researcher combine the scans and techniques from different sources so as to use the strength of one technique to make up for the shortcomings of the other  Neuroimaging data can tell us that a brain area’s activity is correlated with a particular function, but we need to ask whether the brain site plays a role in causing that function  Transcranial magnetic stimulation (TMS) creates a series of strong magnetic pulses at a specific location on the scalp, causing temporary disruption in the brain region directly underneath this scalp area  TMS procedure can provide crucial info about functional role of that brain area Localization of Function  Aims toward figuring out what’s happening where within the brain The Cerebral Cortex  3 categories o Motor areas – contain brain tissue crucial or organizing and controlling body movements o Sensory areas – contain tissue essential for organizing and analyzing info we receive form senses o Association areas Motor Areas  Primary motor projection area – departure point for signals leaving the cortex and controlling muscle movement  Primary sensory projection areas – arrival points for info coming from eyes, ears, and other sense organs Sensory Areas  Each of the sensory areas provides a map of the sensory environment  In each of the sensory maps, the assignment of cortical space is governed by function, not by anatomical proportions  Evidence for contralateral connections comes from the sensory areas Association Areas  Perform tasks of associating simple ideas and sensations in order to form more complex thoughts and behaviors  This terminology is becoming disused  Apraxia = disruptions in the initiation or organization of voluntary action  Agnosia = disruptions in the ability to identify familiar objects; usually affect one modality only  Neglect syndrome= individual ignores half of their visual world  Aphasia= disruption to language capacities  Damage to prefrontal area = problems of planning and implementing strategies Brain Cells Neurons and Glia  Glia: help guide development of nervous system in fetus and young, support repairs if nervous system is damaged, maintain and control flow of nutrient to the neurons  Specialized glia cells provide insulation (myelin)  Cell body = contains nucleus and other elements responsible for metabolic activity  Dendrites = input side of neuron  Axon = output side of the neuron and sends neural impulses to other neurons The Synapse  The gap separating each neuron  All or none law  Each individual response by a neuron is always the same  Transmission across neuronal gap slows down neuronal signal but it is a tiny price to pay for the advantages created by this type of signaling  Communication at the synapse is adjustable; the strength can be altered by experience (learning) The Visual System  Reflected light hits the front surface of the eyeball, passes through cornea and the lens, and then hits the retina (light sensitive tissue that lines the back of the eyeball)  Cornea and lens focus incoming light so that a sharp image is cast onto the retina  Lens is surrounded by adjustable band of muscles  2 types of photoreceptors o Rods – sensitive to low levels of light; color-blind; distinguish among different intensities of light but provide no means of discriminating one hue from another o Cones – need more light to operate; sensitive to color differences; 3 types of cones; allow us to discern fine detail (acuity)  Fovea = center of the retina; here the cones outnumber the rods; this region has greatest acuity  In the visual periphery rods dominate; well out in the periphery there are no cones. This is why we’re better able to see very dim light out of the corner of our eyes Lateral Inhibition  Photoreceptors stimulate bipolar cells  ganglion cells  optic nerve  Optic nerve sends info to lateral geniculate nucleus (LGN)  primary projection area for vision  Cells that link retina to brain are already engaged in the task of analyzing the visual input o Lateral inhibition is a pattern in which cells, when stimulated, inhibit the activity of neighboring cells o Pattern actually leads to stronger responses from cells detecting the edge of a surface than from cells detecting the middle of a surface o This will lead to an exaggerated response along the surface’s edges, making those edges easier to detect = edge enhancements o Edge enhancements help the visual system to discern the shapes contained within the incoming visual info and this occurs at a very early stage of the visual processing Single Neurons and Single-Cell Recording  Single cell recording is a procedure through which investigators can record, moment by moment, the pattern of electrical changes within a single neuron  Electrodes are placed outside a neuron in the animal’s optic nerve or brain and various patterns are flashed on screen  Researchers can then ask which patterns cause neuron to fire etc…  This procedure allows us to define the cell’s receptive field (size and shape of the area in the visual world to which that cell responds) Multiple Types of Receptive Fields  Some neurons seem to function as “dot detectors”. These cells fire at their maximum rate when light is presented in a small, roughly circular area, in a specific position within the field of view  Presentations of light outside of this area cause the cell to fire at less than usual resting rate, so the input must be precisely positioned to make this cell fire o Center-surround cells  Other cells fire at their maximum only when a stimulus containing an edge of just the right orientation within their receptive fields. These cells can be thought of as “edge detectors”  Ex. If a cell’s preference is for horizontal edges, then the cell will still respond to other orientations but will respond less strongly than it does for horizontals o Specifically, the further the edge is from the cell’s preferred orientation, the weaker the firing will be, and edges sharply different from the cell’s preferred orientation will elicit virtually no response Parallel Processing in the Visual System  The visual system relies on a “divide and conquer” strategy  This pattern is plainly evident in Area V1  Visual systems relies on parallel processing – a system in which many different steps are going on simultaneously  Advantages of parallel processing o Speed = brain areas trying to discern the shape of the incoming stimulus don’t need to wait until the motion analysis or the color analysis is complete o Possibility of mutual influences among multiple systems  Neither the shape-analyzing system nor the motion-analyzing system gets priority. The two systems instead work concurrently and negotiate a solution that satisfies both systems  Both types of receptors (rods and cones) function at the same time (another case of parallel processing)  Within optic nerve, there are 2 types of cells o P cells – provide main input for the LGN’s parvocellular cells and appear to be specialized for spatial analysis and the detailed analysis of form o M cells – provide the input for the LGN’s magnocellular cells and are specialized for the detection of motion and the perception of depth  Both these systems are working at the same time (parallel processing)  “what” system = activation from occipital lobe is passed along to the cortex of temporal lobe; plays major role in identification of visual objects  “where” system = activation from the occipital lobe is passed along pathway leading to parietal cortex; performs the function of guiding our action, based on perception of where an object is  Patients with lesions in the “what” system show visual agnosia  Patients with lesions in the “where” system have difficulty in reaching, but no problem in object identification Putting the Pieces Back Together  If multiple brain areas contribute to an overall task, how is their functioning coordinated?  Binding problems – task of reuniting the various elements of a scene, elements that are initially dealt with by different systems in different parts of the brain  What you perceive is not an unordered catalogue of sensory elements. Instead, you perceive a coherent, integrated perceptual world Visual Maps and Firing Synchrony  Various brain areas all have one thing in common: they each keep track of where the target is  The reassembling of the pieces can be done with reference to position  Spatial position is a major organizing theme within all the various brain areas concerned with vision, with each area seeming to provide its own map of the visual world  Evidence is accumulated that the brain also uses a special rhythm to identify which sensory elements belong with which  Neural synchrony: if the neurons detecting a vertical line are firing in synchrony with those signaling movement, then these attributes are registered as belonging to the same object o If they are not in synchrony, the features are not bound together  Attention plays a role in binding together the separate features of a stimulus  When we overload someone’s attention, it is likely to make conjunction errors making mistakes about how the features are bound together (or conjoined)  Studies suggest that synchronized neural firing is observed in an animal’s brain when the animal is attending to a specific stimulus but is not observed in neurons activated by an unattended stimulus  Attention is crucial for the binding problem and it is linked to the neural synchrony that seems to unite a stimulus’s features Chapter 3 FORM PERCEPTION - Vision is dominant sense – reflected in how much brain area is devoted to sight vs. other brain areas - If visual info conflicts with information received from other senses, we usually place trust in vision - How do we see? o Form perception, the process through which you manage to see the basic shape and size of an object o Object recognition, the process through which you identify what the object is Why Is Object Recognition Crucial? - Object recognition is essential wh/e you want to apply your knowledge to the world (and so take action based on what you know) - Crucial for learning - helps to combine different information that you learn on different days about i.e. the same person - Much of knowledge starts: with the recognition and identification of objects in the world Beyond The Information Given th - Early 20 C = Gestalt psychologists – our perception of the visual world is organizes in ways that the stimulus is not o Gestalt is German for “shape” or “form”; the gestalt movement was committed to the view that our theories need to emphasize the organization of the entire shape and not just focus on the shape’s parts o The organization must be contributed by the perceiver; this is why they claimed, the perceptual whole is often different from the sum of its parts o Jerome Bruner “beyond the information given” - The Necker cube – this drawing is an example of a reversible figure –called this b/c people routinely perceive it first one way, and then another o This form van be perceived as a drawing of a cube viewed from above; it can be perceived as a cube viewed from below o Ambiguous figures are often called “reversible” or “bistable” b/c there are two prominent and stable interpretations of the figure - Figure/ground organization, the determination of what is the figure (the depicted object, displayed against a background) and what is the ground - Your perception contains info – about how the form is arranged in depth, or about which part of the form is figure and which is ground – that is not contained within the stimulus itself The Gestalt Principles - Many stimuli (not just the reversible figures) are ambiguous and in need of interpretation - Often don’t detect this ambiguity b/c the interpretation is done so quickly that we don’t notice it - Your perception is guided by principles of proximity and similarity: if , within the visual scene, you see elements that are close to each other, or elements that resemble each other, you assume these elements are parts of the same object - Tend to assume that contours are smooth, not jagged and you avoid interpretations that involve coincidences - Each of us imposes our own interpretations on the perceptual input, but we all tend to impose the same interpretation, b/c we’re all governed by the same rules Organization and Features - WRONG 1. Collect info about the stimulus, so we know i.e. what corners or angle or curves are contained in the input 2. Once we’ve gathered the raw data, we interpret the info, and that when we go beyond the info given – deciding how the form is laid out in depth, deciding what is figure and what is ground - Our interpretations of the input sometimes seem to happen before we start cataloguing the inputs basic gestures not after - The brain areas that analyze a patterns basic features do their work at the same time as areas interact, so that perception of the features is guided by the configurations, and analysis of the configuration is guided by the features - Neither type of processing goes fist – both work together, with the result that the perception that is achieved makes sense at both the large-scale and fine-grained levels OBJECT RECOGNITION Recognition: Some Early Considerations - Recognize objects when your info us partial - Your recognition of various objects, whether print or otherwise, is influenced in important ways by the context in which the objects are encountered - Two types of influences; these influences – coming from the stimulus – are sometimes called stimulus driven but more commonly are termed bottom-up influences - Other influences come from you, rather than the stimulus itself – cases in which you go beyond the information given and more specifically they’re cases in which you supplement the input with your broader knowledge o These influences – relying on your knowledge – are sometimes called knowledge driven or expectation driven, but are more commonly called top-down influences Features - Recognition might begin with the identification of visual features in the input pattern – the vertical lines, curves, diagonals, and so on - You could start assembling the larger units: if you detect a horizontal together with a vertical, you can know you’re looking at a right angle - People are fast and efficient when searching for a target defines by a simple feature - People are slower in searching for a target defines as a combination of features - Damage to the parietal cortex can lead to a disorder – integrative agnosia – people with this appear relatively normal in task requiring them simply to detect particular features in a display o They are impaired in tasks that require them to judge how the features are bound together to form complex objects o Transcranial magnetic stimulation (TMS) is used to disrupt portions of the brain in healthy individuals o Ashbridge found that disruption of the parietal lobe had no impact on performance when participants were searching a display for targets defined by a single feature o The TMS slowed performance when participants were searching for a target defined by a conjunction of features WORD RECOGNITION Factors Influencing Recognition - Tachistoscope, a device specifically designed to present stimuli for precisely controlled amounts of time - Each stimulus is followed by a post-stimulus mask – often just a random jumble of letters, such as “XJDKEL” o The mask serve to interrupt any continued processing that participants might try to do for the stimulus just presented o Researchers can be certain that a stimulus presented for ~20ms is visible for exactly 20ms and no longer - If the stimulus is a word, i.e. we can measure familiarity by literally counting how often that word appears in print, and these counts are excellent predictor of tachistoscopic recognition - Another factor influencing recognition is recency of view, if participants new a word and then, a little later, view it again, they will nd recognste the word much more readily the 2 tind around - The 1 exposure primes the participant for the 2 exposure; more specifically, this is a case of repetition priming The Word-Superiority Effect - Words that are frequently viewed are easier to perceive, as are words that have been viewed recently; words themselves are easier to perceive, as compared to isolated letters - word-superiority effect - This effect is usually demonstrated with a “two-alternative, forced-choice” procedure - In some trials, we might present a single letter i.e. K – followed by a poststimulus mask, and follow that with a question: ‘which of these was in the display – an E or a K’ o In a trial a word would be presented i.e. DARK – followed by a mask, followed by a question which of these was in the display – an E or a K - Accuracy rates are higher in the word condition, and so apparently, recognizing words is easier than recognizing isolated letters - Participants are more accurate in identifying letters if those letters appear within a word, as opposed to letters, appearing all by themselves Degrees of Well-Formedness - Pronounceability: easily pronounceable strings i.e. FIKE do provide a context benefit; if the string is not readily pronounceable i.e. HYZE there’s little or no context benefit o Pronounceable strings are easier to recognize after a brief exposure - Englishness is a good predictor of word recognition: the more English-like the string (measured statistically), the easier it will be to recognize that string, and the greater the context benefit the string will produce Making Errors - Context promotes letter recognition – but only if the context conforms with normal spelling; context that don’t follow normal spelling don’t promote letter recognition - The influence of spelling patterns also emerges in another way: in the mistakes we make - The errors that occur are quite systematic: there is a strong tendency to misread less-common letter sequences as if they were more- common patterns; irregular patterns are misread - These errors can sometimes be large – so that someone shown i.e. TPUM might perceive TRUMPET - Large or small, the errors show the pattern described: misspelled words, partial words, or nonwords are read in a way that brings them into line with normal spelling FEATURE NETS AND WORD RECOGNITION The Design of a Feature Net - A network of detectors, organized in layers, with each subsequent layer concerned with more complex, larger-scale objects - The bottom layer is concerned with features, and that is why networks of this sort are often referred to as feature nets and using the flow of info would be bottom-up – from the lower levels toward the upper levels - At any point in time, each detector in the network has a particular activation level, which reflects the status of the detector at just that moment – roughly, how energized the detector is - When a detector receives some input, its activation level increases o A strong input will increase the activation by a lot and so will a series of weaker inputs - The activation level will eventually reach the detectors response threshold, and at that point the detector will fire – send its signal to the other detectors to which it is connected - If the detector is moderately activated at the start, then only a little input is needed to raise the activation level to threshold, and so it will be easy to make this detector fire o If a detector is not all activated at the start, then a strong input is needed to bring the detector to threshold, and so it will be more difficult to make this detector fire - Activation levels is dependent on principles of recency (when) and frequency (how many times) The Feature Net and Well-Formedness - Adding another layer to the net, a layer filled with detectors for letter combinations – pairs of letters, i.e. or rations of letters - Bigram detectors – detectors of letter pairs – these detectors like all the rest, will be triggered by lower-level detectors and send their output to higher-level detectors o Each bigram detector will start out with a certain activation level, influenced by the frequency with which the detector has fired in the past and by the recency with which it has fired - A strong input will be needed to bring the relevant detectors to threshold, and so the string will be recognized only with difficulty Recovery from Confusion - This confusion is immediately sorted out at the bigram level: all four bigram detectors in this situation are receiving the same input – a strong signal from one of their letters and a weak signal from the other Ambiguous Inputs - pg. 97-98 Recognition Errors – pg. 98-100 Distributed Knowledge - The networks knowledge is not locally represented anywhere; it is not stored in a particular location or built into a specific process - The knowledge about bigram frequencies, is distributed knowledge – it is represented in a fashion that’s distributed across the network and detectable only if we consider how the entire network functions - Most remarkable about the feature net lies in how much can be accomplished with a distributed representation, and this with simple, mechanical elements correctly connected to one another Efficiency versus Accuracy - The network does make mistakes, misreading some inputs and misinterpreting some patterns - These errors are produced by exactly the same mechanisms that are responsible for the networks main advantages – its ability to deal with ambiguous inputs DESCENDENTS OF THE FEATURE NET The McClelland and Rumelhart Model - Designe dintially as an account of how people recognize printed language - One of the early and highly influential models of this sort was proposed by McClelland and Rumelhart - This network is better able to indentify well-formed strings than irregular strings; this net is also more efficient in identifying characters in context as opposed to characters in isolation - These excitatory connections are shown in the figure with arrows; i.e. detection of a T serves to excite the TRIP detector o Other connections are inhibitory, and so i.e. detection of a G deactivates, or inhibits the TRIP detector - Inhibitory connections – a link from one node, or one detector, to another, such that activation of one node decreases the activation level of the other. Often contrasted with excitatory connection - In this model, higher-level detectors (word detectors) can influence the lower-level detectors at any level can also influence other detectors at the same level - Neurons in the eyeballs send activation to the br
More Less
Unlock Document

Only pages 1,2,3,4 are available for preview. Some parts have been intentionally blurred.

Unlock Document
You're Reading a Preview

Unlock to view full version

Unlock Document

Log In


OR

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


OR

By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.


Submit