Unit 3 Key Terms.doc

UNIT 3 ? KEY TERMS PERCEPTUAL DEVELOPMENT Forced-choice preferential looking Teller (1970): Gave infant 2 stimuli, one was plain, one striped Adult observer looks at tiny hole through which they can see infant but not stimuli You can ask the observer where the grating is, they can only watch the infant If the infant shows preference, the adult?s guess is better than chance Collect data from the observers, not infant Results: Takes 6 months to get closer to adult levels Natural adherence of infants to look at patterns rather than plain stuff Adult has forced choice of where the grating is Forced choice: adult forced to choose ; Important method because adult?s performance can be objectively scored as correct or incorrect. Contrast sensitivity of infants Reasons for contrast sensitivity improvement 1. Optics ? Image-forming potions of the eye; infant?s image is not as well focused as adults, causes loss of contrast sensitivity 2. Neural development ? Brain develops more 3. Preferences may not reflect capacity ? The whole method was useless; experiment doesn?t demonstrate infant?s abilities 4. Retinal image motion Retinal image motion Nature-nurture question Effect of restricted rearing on kitten?s visual development (striped drum experiments) Blakemore & others -1970s (kitten in jail picture) Restricted kitten environment for a several weeks then followed with test of visual system Results: Receptive field Showed big change as result of the striped bars Kittens in vertical environments (all simple cells tuned to vertical) Kittens in horizontal environments (simple cells responded to horizontal orientations) Behavior Kitten in horizontal cage tested -- held out horizontal bar and did kitten like things and play with kitten If bar was vertical, kitten acted as if it wasn?t there Conclusion: environment stamps itself on V1 of the kitten Astigmatism Horizontal might be better than vertical, or vice versa Cornea is reason for astigmatism -- cornea is not perfect sphere so it doesn?t show correct image Some orientations come out well, others got blurred Tracked it back to when they were young Even with best optimal correction, the contrast sensitivity of people with astigmatism was never as good as people without it Findings of adults with astigmatism were in track with kitten results ?Carpentered environment? -- most orientations around us are horizontal or vertical Stryker?s experiments Challenged the drum approach Blind experimenter Experimenter doesn?t know the conditions Not told the environment of a given kitten Not told whether it was raised with horizontals or verticals Random sampling Previously -- picked cells in haphazard way; may have found clump of cells with horizontal receptive fields Stryker suggested to make map of V1 and pick cells at random to record Did experiments same as Blakemore; Stryker found nothing Kittens were normal; kittens in vertical and horizontal cells had same cells as normal kittens; no effect of drums at all Perceptual learning in adults Prior to 1990, everyone believed all perceptual learning was done in childhood Karni and Sagi?s sleep deprivation experiments In search task, had random collection of lines, Buried inside 3 collections of lines, either lined up horizontal or vertical Task was to see if you saw horizontal or vertical Results: People got better as they kept doing task Monday -- performance gets better, levels off Tuesday -- took leap from where it was left off Check results with control condition Had them do same task on Tuesday, but stimuli made to fall on different locations of retina (not seen previously) Saw no benefit of overnight breakthrough -- results were not because of familiarity Leap only when tested on same task and same location in retina Brought people to sleep lab REM -- eyes move back and forth, brain activity non-REM -- other stages Deprived subjects of either kind of sleep Woke them up 50-60 times a night Next day results: People with REM sleep showed improvement leap REM-deprived people showed no leap, picked up where left off Learned that whatever you learn through practice during the day, you get more benefit from by sleeping Something happens neurally to strengthen memories to make better performance DEPTH, STEREO AND MORE DEVELOPMENT Corresponding points Images of black dots fall on corresponding points If you take left eye and stick it on top of right eye, line up the foveas of the two, any points that fall one on top of the other are called corresponding points Disparate points = Not corresponding Binocular vision When looking at far target, image of near falls on temporal retinas When looking at near target, image of far falls on nasal retinas If images fall on temporal retina, image is near If images fall on nasal retina, signal is far Position and depth determine the relative positions on the two retinas Binocular neurons are in V1 Will be receiving signals from the two eyes, at different locations Receive signals from nasal retinas, yelling far when it fires Horopter Set of locations (retinal images) that fall on corresponding points Relies on fact that eyes are fixating on black dot Horopter is moving with your fixation If you fix your eyes to the far spot, horopter will change Crossed disparity Object is in front of where you are looking Associated with far Eyes see image on the opposite side Uncrossed disparity Object is behind where you are looking Associated with near Left eye sees image on left side, right eye sees image on right Eyes see image on same side (not crossed) Disparity as a cue to depth Binocular cues tell you relative depth Strongest when dealing with differences that are relatively close --- bad with far distances If you have stereopsis, you can use disparity as cue to depth Demonstration: picture & 3D glasses Green left, red right -- triangle is coming forward Red left, green right -- triangle is going in Depth is fooling your disparity neurons Without glasses, there is just red and green stuff With glasses (filters), you have configuration in your eyes that are not falling on corresponding points Forces images to fall on disparate points Shows how perceptual system uses disparity as cue to depth Random dot stereogram Julesz Strabismus Exotropia or esotropia Misalignment of the eyes (cross-eyed) Disruptive for normal binocular vision because the two eyes do not seem same views of world If misaligned, can?t get same image on the fovea Exotropia: Eyes point outward Can?t get eyes to align on same image -- can?t fall on corresponding points Esotropia: Eyes point inward Amblyopia Unequal vision in the two eyes Ranges from severe (blind in one eye) to moderate One way to deal with it early is to patch the good eye -- weaker eye with less acuity will get better Neural development in normal and binocularly deprived kittens When you raise kittens in environment when the eyes see different, binocular-acy is ruined If you don?t allow binocular view of the world (two eyes seem same seen), then the neural development goes haywire PERCEPTION AND RECOGNITION OF OBJECTS Inferotemporal cortex One of the areas that gets signals from V1 Has large receptive field (3-20 degrees) Most cells prefer moving stimuli Some cells respond to bars, just like simple and complex cells in V1 Some cells respond to cubes (real cubes better than pictures of cubes) Monkey?s paw cell Gross stuck cardboard box in front of monkey?s face -- found that cells responded vigorously When put picture of box, cells didn?t respond as much -- wanted real cube Gross found a cell that wasn?t responding well to anything at all, but when put a hand in front of face, the cells responded well Cells also responded well to toilet brush Single-cell code (Grandmother cell theory) Charlie Gross Studied single cells in inferotemporal cortex Put electrode next to the cell, recorded spontaneous activity, then found out what stimulus reall causes these cells to fire Started with line or ball Found that inferotemporal cortex has large receptive field (3-20 degrees) Most cells prefer moving stimuli Some cells respond to bars, just like simple and complex cells in V1 Some cells respond to cubes (real cubes better than pictures of cubes) Cells that respond to particular objects -- monkey paws, toothbrushes, etc. Shows that cells in inferotemporal cortex respond to objects Ensemble code (Tanaka) Perception of object cannot be done in one cell There has to be a bunch of cells, each taking care of different components There are things that are not grandmother cells and they are things that are working towards grandmother cells There is convergence -- specific areas for certain things, if they are damaged, there will be deficits Shape primitive (Tanaka) Took cat shape, which worked well, but when simplified, cells still responded Change shape into circles/lines fMRI Functional Magnetic Residence Imagery Takes images of soft tissue body parts Able to pick up magnetic signals from brain (comes from iron in blood), magnetic signals are scanned Areas of brain that are most active will draw circulation of blood Present different objects, record which parts of the brain are active Disadvantages: Blood flow is slow -- Can?t do study of perception in real time Measuring regions, not single cells -- Only clump of cells Kanwisher -- used fMRI to learn about perception Showed observer different shapes Showed objects that are facelike and not facelike -- Preference for facelike stimuli Showed places or scenes -- Cells of ?place? area are active but not other stimuli Showed body parts -- What they have in common is not the physical configuration by the category Kanwisher found that the inferotemporal cortex has special subsections for different types of objects in humans Poggio?s model of object perception and recognition Made computer program to recognize objects -- distinguish cats and dogs He traced what system did stage by stage Started with bent wire (in slide) PART 1: Hierarchical model Simple cells (S1) Sensitive to orientation If receptive field gets hit with preferred orientation, it will fire Pooling from level to level -- can?t pool everything Complex cells (C1) Two simple cells with the same orientation pooling to the same complex cell Composite feature cells (S2) Like end-stopped cells Respond best to angle Complex composite cells (C2) Two end-stopped cells, different location receptive fields again TOP -- View tuned cells PART 2: View-based module + view-tuned calls = object-tuned model (next level) Object tuned objects create a problem -- How does the model explain viewing a cat from every possible angle? He pools again For the cat, there will be a cell which fires for every particular view, which pools together and becomes an object-tuned unit Model is based on two things: Pooling Going from one level to the next, and across viewpoints, until we get particular neurons that will respond to your particular object Learning If you want a cell that responds to cats, you need to remember you need cells that respond to cats only You have to know what connections to make as you pool from level to level View-tuned cell Cells that will fire when an object calls in its receptive fields, but only a certain viewpoint Properties created due to pooling from one stage to another We always see objects from different angles (purse example) In real world perception, the retina is the starting point but can?t be ending point, because whenever you change angles, the retinal image changes too Combines hypercomplex or end-stopped cells View-tuned cells are different cells that respond to wire images seen from different viewpoints Object-tuned cell PARALLEL VISUAL PATHWAYS ?What? system X Ganglion cells --> LGN: Parvocellular layers (P cells) --> V1: Simple & complex cells --> Color / Form Color cells Do not care about the shape, whether moving or not Just respond on the basis of wavelength Their activity appears to contribute to their ability to distinguish color Form Respond based on shape Subdivision even past V1 -- neural centers One of them is the inferotemporal cortex (gets signals from V1) ?Where? system Y Ganglion cells --> LGN: Magnocellular layers (M cells) --> V1: Complex cells --> Motion/Depth Motion Bunch of cells that respond on the basis of the speed or direction of motion Don?t care about shape, orientation, wavelength Respond based on pattern of motion Damages produces deficits in perceiving motion Depth MOTION 5 ways we use motion Perceive moving objects Perceive motion-defined forms Track moving objects Navigate using optical flow Perceive motion-in-depth Motion detector (slide) Two neurons at very top (could be cones), intermediate layer, and main neuron If the neural signals reach shaded cell, and there is signal coming from left and from right, which one arrives first? Signal without the synapse -- synapse delays the signal This means right side signal gets there first In order to get strongest response, signals need to arrive at same time -- give slow side a head start (left) This cell will respond better to motion from left to right (rightward) This is basic idea behind motion detector These cells are the motion opponent cells -- receive one direction of signals from left subunit, the right subunit sends opposite signal If both subunits respond, they cancel each other Motion aftereffect Ability to perceive motion after nothing was moving anymore Comes from your motion detection system If you make left subunit tired, you get preference to the right. Counter-rotating spirals illusion Spiral pattern and then look at display of multi-colored chips that appeared to be moving in all directions Generally studied with stripes moving in one direction, after effect is seen moving in other direction Motion signals are generated by our nervous system Direction selectivity Cells responded to rightward or leftward moving stripes MT (Motion area) Contains lots of directional sensitive cells Doesn?t care about anything except direction of motion Optic flow If you have structure in the environment, as soon as you move forward, all images on the ground go streaming by Pattern of the stream of motion depends on where you are hitting it Use this motion to guide ourselves to a target Motion-defined-form Object moving, background not moving Object stands out from background Important when we have nothing else to go on besides motion (i.e. camouflage) Motion-in-depth - Motion of objects coming towards you Martin Regan Interested in real world tasks, particularly dangerous tasks -- game of cricket Cricket ball coming at 90 mph, batsman has little time to react Concluded that visual system has a lot of specialized mechanisms to deal with motion-in-depth (slide of eyeballs) We have neurons sensitive to just these retinal patterns of motion We have binocular motion detectors, combination of two types of neurons Motion-in-depth while driving Gray & Regan -- used driving simulator How close to do you have to be before starting to pass? Adapted to expansion for several minutes to knock out neurons that estimate time to collision After adaptation, observers got much too close to car in front Smooth pursuit Time to collision How we can estimate when collision is about to occur Detection of the expansion of the visual field As we get closer to objects, the images get larger on the retina Localization Two possible causes of motion on the retina: Motion of the object Motion of the eye Two distinguish, we use outflow vs. inflow theories Helmholtz experiment Passive displacement of the eye Tap on eye, which direction does motion move? usually see opposite motion from the tap direction Motor command sent to the eye muscles Copy of the signal of motor command is also sent to visual system Copy of the signal = outflow signal = corollary discharge = efferent copy Mach?s experiment Immobilized his own eye -- used clay in socket so eye wouldn?t move While making effort not to move eye, still saw visual motion (saw things move) Must be pursuing some object out in the world Skavenski?s experiment Used strings to pull on eye, right or left Gave subjects light to look at while they pulled string Eye is immobilized but still fighting the force of pulling strings Visual world moved in proportion of how much fight used in the pull More tug = more fight = more movement Perceptual localization Matin 1970s Briefly flashed light, told observers to indicate perceptual measure of where it appeared to be If flashed when eyes are in motion, incorrect Errors were consistent with slow outflow signal Motor localization Skavenski 1970s Maybe perceptually we don?t see things in term of conscious perception (which may be the slow part) Redid Matin?s experiment, but instead made them whack the light with the hammer -- always correct If you ask people to show you by motor response, it is much better Motor system knows things that we don?t Outflow For perceptual judgments, outflow signal is slow For motor behavior, outflow signal is fast Proposes that perceptual and motor systems have access to different info and may ?see? different perceptual world Monitoring motor command (efferent copy) Perception system knows what the eye did because it knows what commands were sent Inflow Proprioception = position senses; due to receptors in muscles ATTENTION Passive attention Occurs when you get lost in thought i.e. passing an exit while driving because you were thinking about something else Active attention What happens when you have to sit through a boring lecture Voluntary attention can only be maintained for brief periods of time Keep reminding yourself (actively) to pay attention Effects of attention on perception Knock on door is so faint but you only hear it because you are paying attention for it Does it actually make knock louder? -- no because that would mean our perceptual system is distorting incoming info just because we are paying attention to it Effects of attention on memory We remember things better when we pay attention to them Effects of attention on reaction time The time take to react to a new event will be shorter if you are paying attention Attention to objects vs. attention to ideas Cuing experiments Cuing: get advanced info about what is happening; clue Michael Posner Have observer look at central fixation target (asterisk), then a stimulus test probe(red circle) appears, hit key on which side it?s on when it appears Event is detected faster and better at cued location Cue changed criteria for making a decision -- detection vs. decision Dosher & Lu, 2000 Orientation discrimination forced choice task Cue vs No cue Presented cue in display (tiny arrow) Display has 4 tiny sinusoidal gratings End of trial, people asked to report tilt (left or right) of one of them Control cases -- no cue presented Other cases -- cue presented to that subjects know in advance which one they have to report about Better with cue that with no cue Cues improve orientation discrimination Ball & Sekuler, 1981 Cues improve detection of moving patterns All dots move in same direction Display contains bunch of dots that are in motion, all moving in same direction Will measure the minimum contrast (contrast threshold) for dot detection Cue vs. No cue (cue is dealing just with motion detection) Having the cue results in lower contrast threshold for detecting the dots If you get advanced left signal, observer tells brain and motion detecting cells that left cells need to be awake and others be quiet and when stimulus comes, you get big firing of the motion detectors that you need Valid cues Cue given gives you the right answer at least 80% of the time Invalid cues Cue given doesn?t give you right answer at least 80& of the time Peripheral cues If you present cue on one side, and it is valid trial, the test probe really appears on that side, reaction time will be shorter Symbolic cues Arrow points to right, cue is invalid, test probe appears on other side of arrow Attention sensitivity Contrast sensitivity Attention in V1 Attention in area MT (motion area) Attention and counting Change blindness Attention and cell phones AUDITION Compressions Rarefactions Pure tone Sinusoid Sinusoid frequency Cycles per second Sinusoid amplitude Height of sinusoid Fourier analysis Proved mathematically that complex tones can be made by adding sinusoids together Decibel dB = 20log(p/p0) dB = decibel p = pressure (amplitude) p0 = reference pressure Examples: absolute threshold = 0 dB speech = 50 dB truck = 90 dB jackhammer = 100 dB pain threshold = 140 dB Equal loudness curves (slide) Set tones at same physical intensity Middle tones will sound louder This is a description of human hearing, not characteristics of the wavelengths Increase frequency --> gets quieter Frequency determines pitch of sound Experiment -- can?t hear sound anymore at about 20,000 Hertz Complex tone: Anything that?s not a pure tone Squarewave tone: Add pure tones of different frequencies; turn sinusoid into square shaped graph Outer ear Pinna canal (visible) Auditory canal (leads to ear drum) Middle ear Has malleus, incus, stapes (big, medium, small) -- surface area reduced by 20x Stapes: All energy focused in small stapes More impact against membrane when trying to move the liquid Avoids energy loss Inner ear Has cochlea Coiled tube of mostly fluids and neurons Basilar membrane Runs across cochlea Has hair cells which are receptors for hearing (analogous to rods and cones) where neural signals begin Inner hair cells generate neural signals to the neurons of auditory nerve then up to the brain As fluid moves in the cochlea, the basilar membrane vibrates, then hair moves right and left and stretches membrane enough to let particles through This accomplishes transduction: sound --> electrical signal Near stapes -- stiff, best for high frequencies (max vibration) Near apex -- loose, best for low frequencies Tectorial membrane Auditory nerve Traveling wave Place code For a given frequency, you get displaced frequency of a large region Varies with place along the membrane Distinguish frequency tones depending on what cells are firing Near apex -- low frequency Near stapes -- high frequency Temporal code/Phase locking Bursts occur at particular place in the tone Go with frequency of the tone Good for low frequencies -- under 500 Can?t fire fast enough for high frequencies Volley principle Each neuron fires on their ?favorite? time of tone, skips cycles When combined, it is complete Speech spectogram Pick out patterns of speech Phoneme Context effects Categorical perception of speech SKETCHES 1.  The sketch of the two eyes showing the position on each retina of a fixation target and  objects located in front of or behind the fixation target. 2.  The neural wiring diagram of the motion detector.