Allgemeine Psychologie

Introduction to Cognitive Psychology

What is the mind?
- memory, attention, reasoning, problem solving, decision making, language, perception
- creating representations of the world so we can act within it and achieve our goals
- complicated processes involved in easy day actions

Cognitive Science
- interdisciplinary study of the mind = trying to understand the inner workings of the mind
- (Psychology, Computer Science, Linguistics, Neuroscience, Philosophy, Physics and Biology)
- cognitive psychology = study how the mind controls memory, attention, reasoning…

Philosophical antecedents
- rationalist: acquire knowledge through thinking and logical analysis
- empiricist: acquire knowledge via empirical evidence

First cognitive psychologists (~1868 - ~1900)

Donders (reaction time), Wundt (scientific psychology laboratory), Ebbinghaus (forgetting curve)
Donders:
- How long does it take to make a decision?
- Inferring mental responses from the participants behavior
- mental chronometry: measuring cognitive processes
- reaction time (RT) experiment: measure interval between stimulus presentation and response
- Simple RT task: react to presented stimulus
- Choice RT task: two buttons for two different stimuli, making a choice dependent from presented stimulus
- Choice RT – Simple RT = time to make a decision ~100ms (1 sec / 10)
- Assumptions:
- mental operations occur in discrete, independent with choice
stages, that can be isolated stimulus detection + stimulus identification +
- each stage takes certain time to complete response organisation
- can add / delete stages
- Criticism:
- operations could work parallel Without choice
- decisions are not always based on motive / reason stimulus detection + response organisation
- time varies from person to person / situation to situation

Behaviorism (1913 - 1938)

Watson (behaviorism), Skinner (operant conditioning), Pavlov
- abandoning the study of the mind
- relation between behavior and environment
Watson:
- proposed new approach called behaviorism
- eliminate the mind as topic of study and instead, analyze directly observable behavior
“Little Albert”-Experiment:
- classical conditioning of fear
- 9 months old became frightened by a rat after loud noise was paired with presentation of rat
- behavior can be analyzed without any reference to the mind
- how does pairing one stimulus with another affect behavior?

Decline of Behaviorism and the rebirth of the study of the mind (1948)
Tolman:
- one of first cognitive psychologists, because used behavior to infer mental processes
Mouse in cross-shaped maze:
1. mouse initially explores maze
2. learns to turn right to find food at B, when it starts in A
3. food placed at C, mouse turns left to reach B
- Behaviorism predicts that mouse learned to turn right to find food
- Tolman believed mouse created a cognitive map of maze
O’Keefe (1971) and Moser (2005):
- place cells and grid cells provide map, firing when mouse walks through particular loction
- even in dream firing while going through certain mace again

Cognitive Revolution
(1953 Cherry → first commercially digital computer → conferences → Broadbent)
Shift from behaviorist’s stimulus-response relationships to approach that tries to explain behavior in terms of the
mind
Information processing approach (IP):
- a way to study the mind created from insights associated with the digital computer
- early computers processed information in stages
- flow diagram for early computer:
input → input processor → memory unit → arithmetic unit → output
- topics of research:
- how much information can the mind absorb?
- attend to just some of the incoming information?
- questions to consider:
 - how are humans and computers similar?
- how do they differ?
- what are the limitations of this analogy?
~ history of events led to state in humans; AI is put in state
~ similar in simple tasks and both need hardware
~ Computers always same reaction, humans different reaction based on how feel / what’s going on in mind
~ mimicking humans with computers on two different ways:
- mimicking behavior with available technicalities
- mimic neurons firing / not firing with 1 and 0
Cherry (1953):
dichotic listening:
- attention experiment
- present message A in left ear
- present message B in right ear
- to ensure attention, shadow one message
Broadbent (1958):
(siehe early vs late selection models)
filter model as flow diagram
- representing what happens as a person directs attention to one stimulus
- unattended message does not pass through filter
input → → → Sensory Memory → → → Filter → detector → to memory (cf. Chapter “Attention”)

Research
Basic research → important applications
applied research → more basic understanding of cognition
methods:
- self reports
- case studies
- psychobiological research
- naturalistic observation
- computer simulations and AI
- controlled experiments
in an experiment:
- manipulate the independent variable (female / male, noisy / silent room, slept before / not slept before)
- create experimental group
- create control group
- randomly assign participants
- measure the dependent variable
- same for all groups
- control all other variables
- prevent confounds
goals of cognitive research:
- data gathering
- data analysis
- theory development
- hypothesis formation
- hypothesis testing
- application to real world
researching the mind:
- Behavior approach measures relationship between stimuli and behavior
- to understand complex cognitive behaviors:
- measure observable behavior
- make inferences about underlying cognitive activity
- consider what this behavior says about how mind works
- Physiological approach measures relationship between physiology and behavior
- both contribute to our understanding of cognition
Behavior approach:
Muller and Pilzecker (1900)
- 2 groups, 2 lists of words
- independent variable:
- time between first and second list (immediately after first or six minute delay)
- dependent variable:
- memory (recall) for first list
= better performing in delay-group, because memory consolidation was not disrupted
→ memory for recent events fragile
→ if processing disrupted, recent memories can fail to be consolidated
→ new information can interfere with memory consolidation
Gais et al. (2007)
effect of sleep on memory consolidation (just one list)
- independent variable:
- learn list shortly before sleep
- many hours before sleep
- dependent variable:
- memory (forgetting) measured two days later
= better performing in sleep group → memory is consolidated in sleep
Physiological approach:
Gais et al. (2007)
effect of sleep on memory consolidation
- brain activity at encoding and retrieval measured with brain imaging (fMRI)
→ differential brain activity between groups (during word recall more areas functionally related to hippocampus are
active)
- brain imaging techniques to look at areas that are known to be active during memory consolidation: hippocampus
- patient H.M. damaged hippocampus – no new declarative memory, only procedural (golf)

Neural Basis of Perception

Building blocks of nervous system
- Neurons: cells specialized to receive and transmit information in the nervous system
- euch neuron has cell body, axon and dendrites
- they receive, direct and transmit information
How neurons communicate
- Synapse: space between axon of one neuron and dendrite of another
- when action potential reaches end of axon, synaptic vesicles open and release chemical neurotransmitters
- Neurotransmitters cross the synapse and bind with the receiving dendrites

Visual System:
look at an object → light reflected by it enters eye and image of the object is focused onto our Retina = first instance
of a representation

Components of eye:
- Cornea: transparent window into the eyeball
- Aqueous humor: water fluid in the anterior chamber
- Lens: lens inside eyeball allows changing focus
connected through muscles
- Pupil: dark circular opening at centre of iris in the eye, where light
enters eye
- Fovea: centre of Retina
cones very dense → good visual acuity
- Blind spot:
- small spot beside the fovea where optic nerve leaves eye
- no photoreceptors
- goes unnoticed because:
1. away from fovea
2. two eyes compensate for each other
3. brain fills in missing information

- Retina:
- part of brain that sits in eyes
- light-sensitive membrane at back of the eye that contains rods and
cones
(photoreceptors) (→ Duplex Retina) which receive image from the
lens
- photoreceptors transduce (convert) light into neural signal
- neural signal reaches ganglion cells via bipolar and amacrine cells
- ganglion cells transmit signal to brain via axons that emerge together
as
optic nerve
- photoreceptors need lots of blood and nutrition → they are closer to where
blood runs and system is built reverted. Ganglion, amacrine and bipolar
cells
translucent
- Rods and Cones:
- capturing a photon: when light hits eye, process of photoactivation begins
- have special molecules that contain protein called an opsin, that changes
shape, when struck by photon
- have inner and outer segment and terminal

light hits rods/cones → Photoactivation + Transduction → bipolar cells → (amacrine cells, not part of vertical pathway,
but horizontal) → ganglion cells → axons of ganglion cells = optic nerve → brain

Rods Cones

- specialized for night vision = scotopic vision - specialized for daylight vision = photopic
vision
- respond well in low lighting - respond well with lots of light
- do not process color - fine visual acuity and color
- no rods in centre of retina (fovea), most rods to the left - many cones exactly in centre of
retina, nearly no
and right of fovea cones beside fovea

Dark and Light Adaption

- eye can adjust to a wide range of luminance levels
- two mechanisms for dark and light adaption
- pupil dilation: no light → larger pupil / much light → smaller pupil
- photoreceptors: no light → rods are more active / much light → cones are more active
- Rod-Cone break is moment, more rods are active than cones → seeing in dark
Convergence of information via bipolar cells
- convergence: one bipolar cell receives information from more than one photoreceptor
- rods: high convergence
- bipolar cell gathers much information from neighboring photoreceptors and reconciles (abgleichen) all information
- much information joined → good for perceiving motion, but no high acuity
- cones: small convergence
- bipolar cell receives information from very few cones
→ high visual acuity

Ganglion Cells (On/Off receptive fields) (one cell has one receptive field)
- a ganglion cell has a baseline activity
- receptive fields consist of on- and off-areas
- light falling on receptive field changes firing frequency
- when light falls on on-area: firing increases
- when light falls on off-area: firing decreases
- ganglion cell is either
- on-center, off surround cell more sensitive to bright dots
- off-center, on surround cell more sensitive to dark dots
- ganglion cell takes into account how much light falls on to which area and changes firing rate accordingly
- ganglion cells not affected from dim or luminated surrounding

Lateral Inhibition
- excitation of one neuron (one gc) inhibits activity of neighboring neurons
- e.g. hermann-grid illusion: from four sides inhibition (the further away with focus, the larger the field)
- important for perceiving contrasts
- a grey area surrounded by a white area looks darker than grey area surrounded by black area
- doesn’t work if you focus on the spot
- more and smaller receptive fields in fovea → less information converges

Hermann Grid
- illusion disappears, when lines wavy despite distortions produced by curvy lines do not change weighted sum of
perceptive field
- main cause of illusion is straightness of edges of grid lines → illusion is based in striate cortex

Receptive Fields in Striate Cortex (aka Visual Cortex)

- everything that has to do with orientation and straightness is in striate cortex or further up in processing
- many cortical cells respond well to moving lines, bars, edges, gratings or directions of moving
- they are called feature detectors
- Hubel & Wiesel (1965):
- simple cells: neurons that respond best to bars of light of a particular orientation
- complex cells: neurons that respond best to bars of light with a particular orientation and a specific
length
Columns and Hypercolumns
- column: vertical arrangement of neurons; within a column neurons have the same orientation tuning
- hypercolumn: three-dimensional arrangement of neurons, consisting of many columns
- Hubel & Wiesel found systematic progressive change preferred orientation in hypercolumns

Selective Adaption
- method of adaption: less response of a sense organ to a sustained (aufrechterhaltenen) stimulus
- important method for deactivating groups of neurons → “Psychologist’s Electrode”
- if presented with a stimulus for an extended period of time, the neuron adapts to it and stops responding
→ neuron gets fatigue
- idea: human visual system contains individual neurons selective for different orientations
- e.g. The Tilt Aftereffect
- perceptual illusion of tilt (Neigung), provided by adapting to a pattern of a given orientation (tilt)
- selective neurons for a specific orientation are knocked out, if a line of this orientation is present for an extended
time
- when a straight line is presented afterwards, it seems to lean to the other side
- e.g. The Motion Aftereffect
- perceptual illusion of movement in the opposite direction
- selective neurons for moving in a specific direction are knocked out, if a stimulus moving in this direction is present
for an extended time
- when a not-moving stimulus is presented afterwards, it seems to move in opposite direction
- e.g. Face Aftereffect
- after looking at male face, female face looks more female and vice versa

From low-level to high-level areas

- right sides of both eyes pass information on to the left hemisphere of
brain and vice versa
- Optic Chiasm (Sehkreuzung):
- nerve chords from both eyes cross to the respective side
- information from both eyes is exchanged
- visual area in the brain: Occipital lobe

Localization of function

- Occipital Lobe: vision

- Parietal Lobe: touch, temperature, pain
- Temporal Lobe: hearing, taste, smell
- Frontal Lobe: coordination of information received from all
senses

Synaesthesia
- neurological condition in which stimulation of one sensory or cognitive pathway leads to automatic, involuntary
experiences in a second sensory or cognitive pathway
- grapheme-color synaesthesia: letters or numbers perceived as inherently (von Natur aus) colored

Stroop-Effect
- color words are printed in another color
- task: name the color, in which the word is printed
- reaction time is longer for incongruent combinations
- interference effect suggests that we’re not in full control of what we pay attention to

Synaesthesia and Stroop-Effect

- number printed in wrong color
- task: name the color
- reaction time is longer for an incongruent combination

Perception
What is Perception?
- Process of recognizing, organizing and interpreting information from the senses
- not an exact copy of the world
- based on past experience and expectations

Definitions
- Sensation: absorbing raw energy (e.g. sound waves) through sensory organs
- Transduction: conversion of this energy to neural signals
- Attention: concentration of mental energy to process incoming information
- Perception: selecting, organizing and interpreting these signals

From Edges to Objects

Approaches to understand Perception
Bottom-Up processing
- Direct perception theory
- perception starts with the senses; is determined by stimuli in the environment
- incoming raw data, energy registering on the receptors
- parts are identified and put together, then recognition occurs
Biederman: Recognition-by-components theory (RBC) (1987)
- perceiving objects by perceiving elementary features
- Geons:
- three-dimensional volumes
- objects are recognized when enough information is available (but top-down knowledge helps)
- discriminability: can be distinguished from other geons from almost all viewpoints
- resistance to visual noise
- 36 have been identified
Top-Down processing
- Constructive perception theory
- perception starts with the brain, a person’s knowledge, experience, expectations
- people actively construct perception, using information based on experience
- involves making inferences based on context, guessing from experience and basing one perception on another
- happens quickly, automatically
Helmholtz: Theory of Unconscious Inference
- some perceptions are result of unconscious assumptions we make about environment
- we infer much of what we know about the world
- Likelihood-principle: we perceive the world in the way it is most likely, based on past experiences
- e.g. Ebbinghaus’ size illusion: central circle looks bigger, when surrounded by small circles than by big circles
- e.g. Context Effects on Perception: T/-\E C/-\T

Bottom-Up + Top-Down
- Perception of an object is based on:
- Signals representing the object and signals representing other aspects of the environment
- AND Feedback signals representing prior knowledge or expectations

Perceptual Organization
- Middle vision
- loosely defined stage of visual processing
- after basic features have been extracted from image
- before object recognition and scene understanding
- involves perception of edges and surfaces
- determines which regions of an image should be grouped together into objects
- e.g. Kaniszsa Figure and illusory Contours (arrow made up from figures which have spared out parts)
- old view - structuralism
- e.g. Wilhelm Wundt: “Perceptions are the sum of atoms of sensation”
- new view – Gestalt school (>1900)
- Wertheimer, Köhler, Koffka: “The perceptual whole is more than the sum of its sensory parts”
- mind groups patterns according to rules of perceptual organization → Grouping principles
- e.g. Wertheimer’s Phi-Phenomenon: MOVEMENT - circles disappear very quickly → seems like something moves
through them

Gestalt Laws – Laws of Perceptual Organization

- work as heuristics, don’t always result in correct perception
- often provide good estimates/best-guess solutions about properties of the environment
- reflect experience, used unconsciously
- often correct and fast (algorithms are slow)
- Prägnanz: laws of simplicity, every pattern seen in simplest possible way
e.g. Olympic rings as rings
- Similarity: similar things grouped together, similarity in lightness, shape, color, size, orientation
e.g. lines of squares and circles (squares together, circles together) or dark and light circles
oOoOoO
oOoOoO
oOoOoO
oOoOoO
oOoOoO
- Good continuation: connected points result in smoothly curving lines, points seen as belonging together
e.g. tangled cables
- Common Fate: things moving in the same direction
- Proximity (Nähe): things that are near each other are grouped together
e.g.
ooooo
ooooo
ooooo
ooooo
ooooo
- Occlusion(Verdeckung): when large object is partially covered by smaller, we see large one behind small
- Familiarity: things form groups, if groups appear familiar or meaningful
e.g. faces out of stones
- Smallness, etc…

Perceptual Illusions
- compelling but incorrect perceptions
- mostly because perceptual constancies (unconscious heuristics / Gestalt Laws) that normally help perceive more
accurately
- Taking physical and semantic regularities into account (e.g. picture of street scene with kitchen cupboards as houses)
- Monocular Depth Cues (only require one eye):
- patterns of light and shadow
- linear perspective: perception that parallel lines converge in the distance
- relative size: if two objects are of similar size, the one that looks smaller will be judged as further away
- interposition: objects closer may cut off part of our view of more distance objects
- height in horizontal plane
- texture
- clarity: clear objects are judged to be closer
- e.g. haze or aerial perspective (more pale trees), the inverse problem (snow-picture) (shadows: light from above
heuristics)
- Light-from-above heuristic:
- light usually comes from above, so shadows provide specific information about depth and distance
- Perceiving Size:
- Perceived size is a function of bottom-up AND top-down processing
- bottom-up processing:
- size of image on the retina
- top-down:
- perceived distance of the object
- size of the object relative to other objects in environment
- e.g. Ames Room

Basic Principles of Color Perception

- Color is not a physical property but a psychophysical
- Univariance: one type of photoreceptor’s response is the same for an infinite set of different wavelength-intensity
combinations
- Trichromacy (Young-Helmholtz-Theory): color of any
light is defined by relationship of three receptor-types
- Metamers: different mixtures of wavelengths look
identical despite physical differences (NO watercolors)
- Photopic: light intensities bright enough to stimulate
cones and saturate rods (sunlight, bright indoor-light)
- Scotopic: light intensities bright enough to stimulate rods
but too dim to stimulate cones
- Rods:
- all rods contain same photopigment molecule: Rhodopsin
- all rods have same sensitivity to various wavelengths of
light
- therefore, rods: univariance
- Cones:
- S-Cones sensitive to short wavelengths (blue) (least
frequent)
- M-Cones sensitive to middle wavelength (green)
- L-Cones sensitive to long wavelength (red)
- Color space:
three-dimensional space that describes all colors. Several possible spaces
- RGB: Defined by output of long, medium, short wavelengths lights
- HSB: Defined by hue, saturation and brightness
- Hue: chromatic color aspect of light (Farbton)
- Saturation: chromatic strength of hue
- Brightness: distance from black in color space
- Opponent Processes
Lateral geniculate nucleus (LGN) (Teil des Zwischenhirnes) has cells that maximally stimulated by spots of light
- Visual Pathway stops in LGN on way from retina to striate cortex
- LGN cells have receptive fields with center-surround organization
Color-Opponent cell: Neuron whose output is based on difference
between
sets of cones
- In LGN there are color-opponent cells with center-surround organisation

Opponent Color theory:

- perception of color is based on output of three mechanisms, each of them
based on opponency between two colors:
red-green, blue-yellow and black-white
- Ewald Hering (~1900): some color combinations are legal,
while others are not:
- bluish green, reddish yellow (orange), bluish red (purple)
- no reddish green, no bluish yellow
- afterimages: Visual image seen after a stimulus has been removed
- negative afterimages:
- light stimuli → dark afterimage
- color stimuli → complementary color afterimage

Color Perception
- Color constancy:
- intelligent guesses about illuminant
 - assumptions about light sources and surfaces
- e.g. color patches: green patch and grey patch (← excites s-, m-, and l-cones equally)
illuminant changed to reddish light: green patch (← excites cones equally) still looks green, grey looks grey
because red light is taken into account → illuminant is discounted
- visual system’s best guess about what the world really looks like
- exemplary illusions: chess-board with shadow, “the dress”
Object Perception
- How does visual system/brain move from points of light to whole entities, like houses?
- How recognize completely different looking images e.g. as houses?
- How recognize same object from different viewpoints as same? Perceptual constancy

Naive template theory

- recognizing an object by matching every pixel/low-level feature to a representation in memory
- lock and key template idea
- Problem:
- we would need too many templates to recognize every object
- the same thing (e.g. house) can look completely different
- we can also recognize things we have never seen before this way
- it is a learning thing. not born with 50 templates for house, but over the time kind of blurry template as prototypical/
averaging from all seen instances etc. shapes (see also: blurry pictures by Antonio Torralba)
(cf. Chapter “knowledge” - exemplar and prototype approach)
- if shown an atypical boat, time to recognize is longer (told by Vo in lecture)
Extrastriate Cortex
- beyond striate cortex

What-Pathway = Ventral Pathway Where-Pathway = Dorsal Pathway

- identifying objects, recognition - identifying an object’s location,

(no matter where they are) used for guided behavior
- captures fine details - sensitive to motion
- works slow - fast processing
- conscious awareness (bewusstes Wahrnehmen) - neurons respond based on where visual attention is
and Interactions with long term memories allocated
- concerned with the location of objects but not their
functions

Disruption of the “what”-pathway = Visual Agnosia Disruption of the “where”-pathway = Optic Ataxia
- loss of ability to recognize objects / name object - inability to use vision to guide movement
- preserved ability to navigate, reach - inability to reach for an object
- bilateral damage to occipital/temporal lobe
- Balint’s Syndrome
Prosopagnosia (specific to face recognition) - inability to navigate, reach (optic ataxia)
- inability to recognize faces, including one’s own - preserved ability to recognize objects
- cannot recognize person from face - damage to parietal areas
- can recognize objects - inability to perceive the visual field as a whole
- Can discriminate whether two faces are same or different seeing only one object at a time (Simultanagnosia)
- (Nancy Kanwisher's TED talk)

Simultanagnosia
- normal visual fields, yet act blind
- perceives only one stimulus at a time (single word or object)
- if more than one stimulus presented → don’t recognize anything

Dissociation Logic (Perception and Action) (dissociation = Aufspaltung)

- logically deduce conclusions from malfunctions
- damage to different brain areas cause different deficits → conclusion: specific area necessary for specific function
- brain ablation method = damage specific brain areas:
- controlled damage → clear conclusions can be drawn
- Single dissociation: one function lost, another remains
→ what and where rely on different mechanisms, although they may not operate totally independent of one another
- Double dissociation: two individuals with different damage and opposite deficits
→ therefore, what and where streams must have different mechanisms AND operate independently of one another

Face Recognition (Special Case of Object Recognition)

- Part of Inferior Temporal (IT) Cortex (inferior = untergeordnet):
Fusiform Face Area
- Cells firing strongly, when shown faces
(e.g. monkey: face of monkey > face of human >> face of human without eyes,
cell-firing for smiley just a bit above random-noise (baseline))
- Jennifer Aniston-Cell just firing for Jennifer Aniston (but not with Brad Pitt and not for other female faces)
- Greebles (Gauthier, 1999): Participants see Greebles. Before Training, almost no response in FFA, after Training
much increased firing for Greebles, while firing for humans slightly decreases (just a bit more than for Greebles)
- FFA also reacts to changed/incomplete pictures of faces:
- Face Inversion Effect: face inverted and shown upside-down
- Moony Faces: faces as pattern of black and white

More specialized brain regions

- LO: Lateral Occipital complex: OBJECTS
- PPA: Parahippocampal Place Area: SCENES
- EBA: Extrastriate Body Area: BODY PARTS
- VWFM: Visual Word Form Area: WORDS

Environmantal Influences on Neurons

- some neurons respond best to things that occur regularly in the environment
- neurons become tuned to respond best to what we commonly experience:
- horizontals and verticals
- experience-dependent plasticity (e.g. Greebles)
- e.g. cat raised to see horizontal lines → cells don’t fire for vertical ones

Mirror Neurons
- located in premotor cortex
- activated when trying to understand another person’s action
- important for imitation → learning things
- important for empathy
- neurons firing (almost) same when performing an act as when someone else performing this act
- e.g. premotor-cells of monkey firing lot when experimenter grabs something. firing as well (more) when monkey
grabbing. Firing very little, when experimenter grabs with pliers

Objects in Scenes - Objects in context

e.g. Antonio Torralbas pictures: prototypical objects AND prototypical backgrounds
- can we identify an object faster if the scene is consistent?
Mutual Interference (consistency) (Wechselseitige Beeinträchtigung)
- consistence in scene: better recognition for objects and background
- inconsistent scene: worse recognition for objects and backgrounds (but better for objects than for backgrounds)
e.g. Biederman (1981): picture of hydrant on mail box
- semantic consistency: a scene is glimpsed briefly and affects both, object and background perception
→ objects and their settings are processed interactively and not in isolation

Scene Perception
(- easier with color)
- global approach: identify entire scene at once, without identifying individual objects
→ ‘Gist’ (Kernaussage) or ‘Gestalt’ processing (cf.: more than the sum of...)
- in a glance, remembering meaning of an image and its global layout, while some objects/details forgotten
- local approach: identify individual objects in a scene and use identities to determine what the scene is
- ensemble statistics (ensemble = das Ganze)
- knowledge regarding the properties (like orientation or color) of a group of objects or a region in a scene
- e.g. briefly shown image:
II II --- -_
I II I - _ -_
- can tell very accurately orientation of red and blue bars. This speaks more for global approach
- seems like we’re doing this in scenes, too

Signal Detection Theory

- reasoning and decision making: always some uncertainty left
- SDT is precise language and graphic notation for analyzing decision making in presence of uncertainty
- two main determinants:
Discriminability = d’ (given by context)
Decision Criterion = β (sometimes c) (changed actively, attention)
- task: decide if stimulus is present or not in a “noisy” environment (inner distraction is also noise)
- correct decision: hit / correct rejection
- wrong decision: miss / false alarm
- probability for hit, miss, correct rejection, false alarm changes
dependent of discriminability and decision criterion
- real world applications: cancer- or baggage-screening
- low prevalence (geringes Auftreten):
- overall appearance of target influences criterion
- target is rare → decision criterion β shifted to more
conservative
 - little less false alarms, substantially more misses

Changing in discriminability Changing decision criterion

- higher d’ (less distraction, noise) - conservative β (prevent

false alarms):
= high signal-to-noise ratio: - more misses and correct
rejections
- correct decision more probable - less hits and false
alarms
- wrong decision less probable
- liberal β (prevent misses):
- low d’ (loud) - more hits and false alarms
= low signal-to-noise ratio: - less misses and correct

Attention
What is Attention?
- mechanisms that selects certain aspects of a scene for further processing while ignoring others
- process of concentrating on specific features of environment or on certain thoughts or activities
(- concentrating mental energy to process incoming information)
- main functions: signal detection and vigilance (Wachsamkeit), search
- selective attention (excluding other features of environment) or divided attention
- limited in space and time

Endogenous Attention:
- top-down
- attentional process which has voluntary aspect and comes from within
- e.g. decide to put attention always to the left

Exogenous Attention:
- bottom-up
- endogenous attention can be overridden by salient (bedeutend) and powerful stimuli generated outside oneself
- e.g. some noise or movement in corner of eye

Stimulus characteristics that affect attention: Personal factors that affect

attention:

- intensity - motives
- novelty - emotions
- movement - interests
- contrast - task
- repetition - vigilance

Attention and Visual Perception

Inattentional Blindness:
- stimulus that is not attended is not perceived, even though might be looked directly at it
- e.g. (Mack & Rock) cross with differently sized arms and with objects between arms.
task: name longer arm, real task in the end: which object? Participants clueless
- real-world
- scanners: slice for slice from top to bottom
- drillers: keeping eyes fixed, waiting for detection of spots
- e.g. over 80% of radiologists did not perceive gorilla in lung

Change Blindness:
- shown two versions of a picture, differences not immediately apparent
- identifying differences requires concentrated attention and search
- e.g. picture where bar in background goes up or some plants are gone

Selective Attention
- ability to focus on one message and ignore all others
- we do not attend to large fraction of information in environment
- filtering out some information and promoting other information for further processing
- Research method: Dichotic Listening
Can we completely filter out unattended message?:
- participants could not report content, but knew there was a message and knew gender of speaker
→ unattended msg is being processed at some level:
- Cocktailparty-effect: 1/3 of participants hear own name or “fire” in unattended message
 - change in gender is noticed
- change in tone is noticed

Early vs. Late Selection

- Does attention filter occur early or late in processing?

Early Selection model:

- Broadbent’s filter model

- Filters message before incoming information is analyzed for meaning

- Sensory Memory holds all incoming information for fraction of a second and transfers all information to next stage
- Filter identifies attended message based on physical characteristics and only passes on attended message
- Detector processes received information to determine higher-level characteristics of the message
- Short-term Memory receives output of detector, holds information for 10-15 sec and may transfer to LTM
- could not explain cocktail-party phenomenon
- could not explain that participants can follow meaningful messages that switch from one ear to another
- could not explain effects of practice like being trained to detect in unattended ear, based on meaning of msg

Intermediate Selection Model:

- Anne Treisman’s attenuation theory

- attended message can be separated from unattended early in information-processing system, but selection can also
occur later
- Attenuator (Abschwächer) analyzes incoming message based on physical characteristics, language and meaning.
attended msg is let through at full strength, while unattended msg is let through at much weaker strength
- Dictionary Unit has activating-threshold (Schwelle) for words, regarding low threshold for common and important
words and high threshold for uncommon words (rutabaga → high threshold, own name → low threshold

Late Selection Model:

- Deutsch & Deutsch (1963): Selection of stimuli for final processing does not occur until after information has been
analyzed for meaning
- McKay (1973):
- attended sentence ambiguous (mehrdeutig)
- unattended biasing word which participants were unaware afterwards
- biasing word affected participants’ choice of which sentences were closest to attended
- (throwing stones at bank, river, money)
→ argued that most info is processed to level of meaning before message to be processed is selected

Early-Late Controversy
- every selection model can be demonstrated under some conditions, depending on task and type of stimulus
- “Cognitive load” plays important role

Attentional Selection in Space

Covert Attention: Attention without eye movement
- wikipedia: covert attention is the act of mentally shifting one's focus without moving one's eyes
- precueing: directing covert attention without moving the eyes
- pioneered by Michael Posner (1980)
- even with fixed eyes participants respond faster to light at expected location
- Endogenous Cueing Task
- location-based: moving attention from one place to another
- boxes to the left and to the right of fixation point
- cue appears, either →, ← or ↔
- afterwards target in one of the boxes appears
- RT is shortest, if target appears in cued box (valid) and longest if in invalid box
→ difference in time is moving attention to the other side
- object-based: attention being directed to one place on an object
- e.g. Egly et al. (1994):
 - 1 sec preview, 100ms cue, 100ms delay
→ object-based attention: shorter RT for same distance in same object
- Exogenous Attentional Capture
- green circles and one square arranged around fixation point
- task: find square
- RT goes up, if color-distractor (one of circles is red)
- set size has almost no impact if just green ones (bigger impact when red is included)
- Simon-Task
- press button: right hand for butterfly on screen, left hand for frog. Target appears on right or left side of fixation point
- reaction times are faster and more accurate, when stimulus occurs in same relative location as response
- even if stimulus location irrelevant to task
- e.g.: when alarm light flashes at left side of dashboard, reaction is faster, when emergent brake also left
→ Simon Effect: implications for human-machine interface design
- Stimulus → Stimulus Identification → Response Selection → Response Execution
- cf. Interference that produces Stroop-Effect
- endogenous and location-based?

Attentional Selection in Time

Attentional Blink (AB)
- short period during which incoming information is not registered, almost like physical eye blink
- information that is attended to, blocks brain for short moment
- some models propose bottleneck: restriction of amount of information that can be processed at once
- Rapid Serial Visual Presentation (RSVP) & Attentional Blink (AB):
- competition for attention over time instead space
- RSVP sequences contain at least 2 targets within a stream of distractors
- SOA: Stimulus onset asynchrony (interval between targets) is varied by placing second target in different serial
position → lag
- targets: e.g. letters, distractors: digits (presentation rate usually 100 ms)
- brief temporal gap in reported perception shortly after attention has been dedicated to target
- if second target falls into gap, it is likely to be missed
→ interference of T1 with T2
- effect max when SOA between T1 and T2 is 200 ms to 400 ms
- how to make sure, it is attentional blink and not perceptional blink?:
- set up experiment: ignore T1
- (use eye-tracker or cross-check with another type of stimulus, so it is not the eyes)

Cognitive Load and (selective) Attention

- Cognitive Load:
- how much of cognitive resources are used to accomplish a task
- high-load: uses almost all; no resources for other tasks
- low-load: uses few; resources for other tasks
- Flanker-Compatibility Task a)
- target is flanked by two other items
- task: press one key, if target is A or B, press another if target is C or D (ignore flankers)
- compatible: (e.g. A flanked by B)
→ low cognitive load, fastest response
- incompatible: (e.g. A flanked by C)
→ high cognitive load, slowest respond
- neutral: (e.g. A flanked by X)
→ intermediate response
- Flanker-Compatibility Task b)
- array of letters
- task: press key, if target is present
- flankers only circles and one distractor (e.g. N) → low work load
- or many distractors (different letters) → high work load, because many potential targets
- same number of flankers: RT longer for high work load
- AND in high-load condition, type of distractor doesn’t affect reaction time because no resources to process distractor

Divided Attention
- naturalistic driving study: video cameras in cars
- risk of accident four times higher, when using cell phone
- Strayer & Johnson (2001):
- simulated driving task
- participants on cell phone missed twice as many red lights and took longer to apply brakes
- same results using hands-free devices

Hemispatial Neglect
- deficit of attention in which one entire half of a visual scene is simply ignored
- massively exaggerated (verstärkt) version of our attentional failures
- often caused by stroke that has interrupted flow of blood to right parietal lobe
- only one half (more and less possible) of visual field is attended to
Attention Deficit Hyperactivity Disorder
- Inattention: easily distracted
- Hyperactivity: “Zappelphillip”
- Impulsivity: seem very impatient (ungeduldig)
- 3 times more common in boys
- 40-50% childhood continue adulthood
- 2-5% of adults have this condition
- treatment: Psychotherapy and Ritalin

Autism Spectrum Disorder (AS)

- serious developmental disorder (Entwicklungsstörung)
- one of major symptoms: withdrawal of contact from other people
- Adults with AS have different patterns of visual attention compared to neuro-typical (NT) in social scenes
- could solve reasoning problems that involve social situations
- could not function when placed in actual social situation
- this may be due to the way they observe what is happening
- autistic observers look to the mouth or the side of the face to assess (beurteilen) emotional reaction, instead of to
eyes
- autistic attention (or lack thereof) may be one way in which they see things differently in environment

Feature Integration Theory (FIT), Anne Treisman

- Preattentive stage:
- automatic, unaware process
- no effort or attention
- object analyzed into features like colors, shapes, orientation, motion, curvature, depth
- Focused attention stage:
- attention plays keyrole
- features are combined/bound
- (location of items also here)
- The Binding Problem:
challenge of binding together different attributes (features) of visual stimuli to perceive unified object
- Illusory Conjunctions:
- combination of features from different stimuli (objects)
- occur because features are “free floating” before focused attention stage
- Treisman & Schmidt:
- presentation of an array of geometrical objects and black numbers
- task 1: concentrate on objects, ignore numbers
focused attention stage (bottom-up processing)
→ shapes and colors correctly paired
- task 2: concentrate on black numbers
only preattentive stage
→ illusory conjunctions occur
- task 3: participants were told, what they would see (e.g. lake, carrot)
→ no illusory conjunctions occur because of top-down processing
- Balint’s Syndrome patient R.M.: high number of illusory conjunctions
- in everyday life: combination of top-down knowledge and bottom-up processing helps to perceive things accurately

Visual Search
- looking for target amongst distractors (car in parking lot, friend in crowd)
- salience (salient = auffällig): Vividness (Klarheit, Lebhaftigkeit) of stimulus relative to neighbors
- distractor: any stimulus other than target
- set size: number of items in visual search display
- the efficiency of VS is quantified as average RT as a function of set size
- measured in terms of search slope ms/item
- the larger the search slope the less efficient the search
- search asymmetry: detecting absence is harder than detecting presence / detecting unusual amongst usual harder
than usual amongst unusual
- harder when not familiar (e.g. chinese letters)

- Feature Search
- search for target defined by single attribute such as salient color or orientation
- mostly efficient (not if e.g. one red 2 in array of red 5, but is this at least feature search?)
- is parallel: multiple stimuli at a time – get whole field and target pops out

- Conjunction Search (conjunction = Verbindung)

- no single feature but the co-occurrence of two or more features defines target
- efficiency heavily depends on set size
- most real searches are conjunction searches
 - often inefficient

Feature vs. Conjunction search

search slope

Guided Search, Jeremy Wolfe

- core idea of guided search: conjunction searches
- set of guiding attributes: orientation, colors, etc.

- Binding Problem
- to recognize objects, we must bind features and to bind features, we must attend to the object
→ so to find an object, we must search

- Compute Bottom-up Guidance

- stimulus-driven
- local salience is computed
- local differences create bottom-up salience
- Bottom-up saliency is often not enough to find target (e.g. colored magnetic field with some green verticals)
(must give weight to what you want)

- Compute Top-down Guidance

- user-driven
- limited set of coarse (grob), categorial features are computed
- weighted sum creates priority map
- (but bottom-up information might interfere with top-down concentration – e.g. picture of Trumpeltier)

- Priority Map
- all else being equal, local salience will be weighted heavily and will attract attention (bottom-up)

- in priority map for conjunction search, for each characteristic (color, orientation) there is a cognitive map, where
all objects are highlighted, that match the criterium
- e.g. one map where all red objects are highlighted, one where all verticals are
- in priority map both characteristic are combined
- only objects that match all criteria are highlighted on priority map
- attentional bottleneck, because attention cannot be at all locations, but attention is guided already in correct
locations
- maybe real objects are just elaborated (ausgearbeitete, vervollkommend) conjunctions of basic features?
- in not-random array (real life scene) there are other mechanisms that help us search – because mostly not looking at
feature driven but semantic driven things (bike in street-scene: not looking for parts of bike, but on street)
- does not explain, why sth. is harder to find in random array

Eye Movement and Attention

- wikipedia: overt attention: selectively attending to an item or location by moving the eyes in that direction
- overt (offenkundige) attention, composed of:
- Saccades: rapid movement of eyes from one place to another
- Fixations: short pauses on points of interest
- studying search in scenes: tracking eye movement
- state of the art: reflection on eye and pupil-movement
- eye movement, because fovea needed for sharp vision
- only possible to see something during eye is moveless

What determines where we look?

Bottom-Up determinants (bestimmende Faktoren) of eye movement

- Stimulus Salience (Auffälligkeiten): Areas that stand out and capture attention
- depends on characteristics of stimulus
- color and motion highly salient
- when lot of cuts on screen: centre of screen
- Itty & Koch: Saliency model (2001)
 - saliency maps for colors, intensity, orientation, other characteristics (motion, etc.)
- are combined to saliency map, where areas are highlighted, that are salient on many individual maps
- predict where people look

Eye movement in real-world tasks

- What guides search in complex environments? (e.g. where is Waldo?)
- bottom-up saliency not enough to explain eye-movement in scenes
- visual world complex but highly rule-governed

Dual-Pathway Model (early vision)

- nonselective pathway: contextual guidance - perceiving whole scene
- selective pathway: feature guidance – bottleneck → binding and recognition (put together to (whole) scene)

Top-down determinants of eye movement

- scene schema/grammar: knowledge about what is contained in typical scenes
- beyond feature guidance:
- Scene Semantics: Knowing what objects should be in scene
- Scene Syntax: Knowing where objects should be in scene
- help guide fixations efficiently from one area to another
- eye movement is also highly determined by the task
- e.g. Yarbus (one of founders of modern eye movement research) (1960s)
- “The unexpected visitor” under different tasks (free examination, ages of people, material circumstances of family...)

Physiology/Neural Basis of Attention

- Attention enhances (erhöht) neural responding
- Attentional processing is distributed across a large number of areas in the brain
- Physiology of Covert Attention
- monkey trained to fixate eyes on dot while peripheral light flashed
- fixation only: monkey was to release bar, when fixation light dimmed
- fixation and covert attention: monkey was to release bar, when peripheral light dimmed
- single cell recordings showed monkey’s response when paying attention to the peripheral light was not caused by
changes of the stimulus on retina, but by the monkey’s attention to light

Attention in Social Situations

- dynamic face viewing:
- social events:
- eye contact: looking at a person’s eyes
- talking: looking at mouth
- head movement: looking at nose (anchor)
- Gaze following: if talking person talks about another person and looks at them, viewer follow their gaze

Memory
rehearsal = einüben
retrieving = abfragen
retention = Beibehaltung

What is Memory?
Processes involved in retaining (behalten), retrieving (abfragen) and using information about stimuli, images, events,
ideas and skills after the original information is no longer present

Modal Model of Memory – Atkinson & Shiffrin (1968)

- Computer as model for human cognition
- Memory as integrated system that processes information
- Acquire, store and retrieve information
- components of memory don’t act in isolation
- control processes exist: active processes that can be
controlled by the person
- rehearsal (einüben)
- strategies used to make a stimulus more memorable
- strategies of attention
- Memory has limited capacity
- limited space
- limited resources
- limited time

Sensory Memory
- short-lived sensory memory registers all/most information that hits our visual receptors
- information decays very quickly
- persistence of vision (perception briefly held in mind): retention (Beibehaltung) of the perception of light
- sparkler’s trail of light
- frames in film
- holds large amount of information for a short period of time
- collects information
- holds information for initial processing
- fills in the blanks
 - keeps beginning of sentence before sentence is finished
Perceptual Span
- researchers were interested in something called the “perceptual span”
- how much information can be gathered in single percept
- first results claimed that people could accurately report about 4.5 items from a brief percept
- Sperling suggested that the 4.5 item limit was imposed not by the capabilities of the perceptual system, but by
observers’ abilities to recall items that had been seen
→ not perception but memory is responsible for 4.5

Measuring capacity and duration of SM – Sperling (1960)

- arrays of letters (3x4) flashed briefly on screen
- whole report: report as many letters as possible
- average 4.5 out of 12 letters
- partial report
- participants heard tone after visual presentation that told them which row of letters to report
- average 3.3 out of 4 letters
- participants could report any of the rows
- delayed partial report
- presentation of acoustic signal delayed for a fraction of a second after the letters were shown
- performance decreased rapidly

→ perceptual span in earlier experiments determined by time needed to report

- duration of information in SM only a few hundred milliseconds up to 1 sec
→ perceptual span not 4.5 items but rather number of items in entire visual field

Short-term Memory
- stores small amount of information for brief duration
- includes both new information received from the sensory stores and information recalled from LTM
- STM is about 15-20 seconds, if no rehearsal (kein einüben)
- Proactive Interference (PI) occurs, when information learned previously interferes with new information
- capacity of short-term memory = digit span: how many digits can person remember
- typical result = 5-8 items
- item not necessarily digit, but chunk
- George Miller: Magical Number 7 (1956):
- STM has key limitation: 7 (+-2)
- number of items matters, their sizes don’t
- if separate items can be grouped into chunks, STM more efficient
- e.g. FB – INB – CC – IAIB – M → FBI – NBC – CIA – IBM
- Chunking: small units can be combined into larger meaningful units
- chunk is collection of elements strongly associated with one another but weakly associated with elements of other
chunks
- top-down knowledge helps build meaningful associations (as in perception)
- e.g. Ericsson (1980) trained college student S.F. to go up from 7 to 79 digits
- chunking into meaningful units: race times, ages and dates creating mnemonics (Eselsbrücke)
→ shows interactions between LTM and STM
- Chase & Simon (1973) tested chess-players to memorize chess pieces on board
- either randomly positioned or for real game
- chess masters and beginners
- chess masters better performance for real game positions
- no advantage for randomly arranged pieces

Limitations of this Model of STM

- STM was mainly described as short-term storage mechanism
- but the role of STM goes beyond storage
- involves also transfer from and to LTM
- dynamic processes need to be considered as well
- this is when Alan Baddeley’s WM Model was born

Working Memory – Alan Baddeley (1974)

- similar concept to STM
- WM: limited capacity system for temporary storage and manipulation of complex tasks like comprehension
(Verständnis), learning and reasoning
- WM differs from STM
- consists of multiple parts (STM is single component)
- is concerned with the processing and manipulation of information that occurs during complex cognition, while STM
only stores Information for brief moment

- Phonological Loop
- verbal and auditory information
- phonological store
- memory store that can retain (beibehalten) speech-based (phonological) information for a short period of time
- unless rehearsed (eingeübt), the traces within the store are assumed (vermutet) to fade and decay within about 2
seconds
- the articulatory control process
- translates visual information into a speech-based code
- deposits it in the phonological store
- it refreshes a trace in the phonological store to shield from decay
- idea that there is a system specialized for language
- evidence comes from findings like
- word-length effect
- phonological similarity effect
- articulatory suppression
Word-length effect
- memory for list of short words is better than for list of long words
- takes longer to rehearse (einüben) long words and produce them during recall
Phonological similarity effect
- sound more important than meaning?
- letters or words that sound similar are confused
- for silently written list, too. reason: written information is transformed into auditory information (articulatory control
process)
- only works if articulatory control process can be performed correctly (no articulatory suppression)

Articulatory suppression
- e.g. saying blablablablabla… while trying to memorize
- prevents from rehearsing (eingeübte) items to be remembered
- reduces memory span
- eliminates word-length effect
- reduces phonological similarity effects for reading words
- prevents articulatory control process from recoding the visual information into phonological information
- experiment reported by Murray:

- articulatory suppression nukes effect. Outcome for both variants with suppression same
→ phonological similarity effect also for written, because articulatory control process converts information into
phonological form. If this is prevented by articulatory suppression, phonological similarity effect removed

- Visuospatial sketch pad

- handles visual information and is involved in visual imagery - the creation of visual images in the mind - in the
absence of a physical visual stimulus

Experiment, Brooks
- visualize capital letter ‘F’, identify corners as ‘in’- or ‘out’-corners
- starting with top left corner
- either naming kind of corner ‘in’ or ‘out’ (phonological)
- or pointing to ‘out’ or ‘in’ (visuospatial)
- speaking easier than pointing
- task to visualize involved visuospatial sketchpad
- verbal response involved phonological loop
- conducting two visuospatial tasks overloaded the visuospatial sketchpad

Conclusions for WM
- results show: if task and response draw on same WM component → performance worse than when distributed
between WM components
- WM is set up to process different types of information simultaneously
- WM has trouble when similar types of information presented at same time

- Central Executive
- component that most strongly differentiates idea of WM from earlier concept of STM
- determines what gets into WM
- determines where information is stored (visual vs. verbal)
- integrates and coordinates information
- Allows ‘work’ on this information: inspection, transformation (e.g. maths), and other cognitive manipulations
- Attention controller: focus, divide, switch attention
- controls suppression of irrelevant information

Limitations of Baddeley’s WM-Model

- WM can hold more information than expected (e.g. sentences longer than 20 words)
- introducing episodic buffer (update of WM 2000)

Episodic Buffer
- auxiliary store (Hilfsspeicher) when primary ones overloaded or disrupted
- backup store that communicates with LTM and WM components
- holds information longer and has greater capacity than phonological loop or
visuospatial sketchpad
WM and the brain
- prefrontal cortex responsible for processing incoming visual and auditory information
- monkeys without prefrontal cortex have difficulty holding information in WM
- Delayed Response Task (Funahashi)
- monkey fixates eyes, stimulus appears in visual field
- single cell recordings: in prefrontal cortex neurons responded when stimulus was flashed in particular location
- when stimulus disappeared, neurons continued firing till monkey moves eyes to location where stimulus was shown
- information remains available via those neurons for as long as they continue firing

- Long-Term Memory
- declarative (explicit): episodic (events, experiences), semantic (facts) – everything that can be spoken out
- non-declarative (implicit): procedural (movements), perceptual (recognizing faces), conditioning (incentive based
training) – hard to verbalize, like how to ride a bike

Learning and Memory (guest lecture by Henning Beck)

Part 1: General features of learning systems

- False Memories
- DRM paradigm (Deese, Roediger, McDermott (mid 1990s)):
- inducing false memories by:
1. presenting list of semantically related words
2. delay time
3. task: recall as many words as possible (works with shown second list or self driven)
~ 70% of participants remember items that were never presented
e.g. [tasty, cake, good, sugar, nice, pastries, bakery, tooth, honey, waffles, eat, milk] → delay →
[sweet, plate, cookies, phenoxyethanole] (remembering all, except the last; none was on list)
→ brain not storing information, but organizing information (“it is about organizing general concepts”)

Where are memories located?

- in a neuronal network: not located anywhere specific, but in a way how neurons interact at certain moment
- like an orchestra: it is the way how musicians interact at certain moment
→ specific single Jennifer Anniston-cell does not exist, but cells are almost the same for specific memory – like fans in
hockey stadium: maybe one gets lost, maybe new one joins
- in this process neurons communicate with each other (via their synapses)
- neurons need to adjust (anpassen) their connections according to the stimulation (by interactions they can change
network)
- What fires together, wires together
- the older the neuron, the lesser the density of synapses
- use it or loose it
- synaptic pruning: approx. 90% of all synapses lost over first two years because of fitting to environment
- if too many connections “it is hard to get through”
- energy is important factor, too

This type of learning has three problems:

1. it is too slow (neurons need minutes or hours to adjust morphologically to input)
- so how are we able to learn new words like ‘selfie’ or ‘brexit’ in seconds?
2. neural networks tend to ‘overfitting’
- if input too similar
- e.g. if european looks at asian korean beauty queens, pictures look almost same
- e.g. amazon algorithm
- how are we able to learn the important information and generalize from a few examples?
3. neural networks are prone (gefährdet) to ‘catastrophic forgetting’
- information can be overwritten by new information if same cells activated
- e.g. a cucumber is edible, green, elongated and floating
a cruise-liner is elongated, floating, heavy and loud
- how do we secure information from being ‘overwritten’ by new information?

Solution: a two-step learning system: cortex and hippocampus

- cortex: slow, steady and robust
- hippocampus: fast, flexible and dynamic
- not place where memory is stored
- is memory organizer
- without it is hard or impossible to get new memories (patient H.M.)
 → time →

- memory replay takes place during the change of different sleep phases (dreaming not necessary for memory
formation)
1. awake: gaining semantic knowledge (e.g. via school, TV, etc. learned stuff about solar system)
2. REM: restructuring semantic knowledge regarding overlapping representations
3. non-REM: extracting gist and building schemas from information in Hippocampus and putting it in Neocortex
4. REM: - hippocampus is blanked and gist of all new information in neocortex, now.
- anatomical pons between new and old knowledge-topics are made and topics are restructured with
connections inbetween them (e.g. gist of new information about solar system and gists of old information
about atoms and concentric circles)
→ Memory replay during sleep does not only improve learning but also gist extraction (and creative problem-solving)

→ Solutions to problems
1. ‘robust neural networks are rather slow’: highly dynamic hippocampus, slower cortex
2. ‘neural networks overfit’: hippocampus deletes its information after sleep
3. ‘catastrophic forgetting’: predictive coding (brain does not learn hard facts, but models)

Part 2: Special flavors of learning

- Schema learning
- schema = cognitive framework that is used to categorize information
- exemplar approach and prototype approach converge to schema
(cf. Chapter “knowledge” - exemplar and prototype approach)

- e.g. Sievers (2019): angry drawings have on average 17-24 corners, sad objects 7-9, independent from culture

Schema learning is using extremes to build an average category:

- children use extreme examples to categorize a schema (based on best representative (cheetahs)),
adults use the average example to build a schema but still take extremes to describe the “best” example

Average Hypothesis
1. a car is a vehicle with an engine and four wheels
2. a car is a vehicle with an engine,four wheels,four doors,a trunk,maximum five seats and six windows to move people
3. a car is a vehicle

Schemas can be built from a single example and transferred to novel situations
- e.g. this is a ‘tufa’. Which ones are also ‘tufas’?

- Inductive Learning (induktiv := vom Einzelnen zum Allgemeinen hinführend)

(- until 18 month babys know on average 50 words. Vocabulary spurt takes place after 20 months of age.
Children can learn 8 novel words per day)

Explicit Encoding
- shown examples of item to learn item (This is Noni Fruit, remember Noni Fruit)
- retention (Beibehaltung) better than via Fast Mapping

Fast Mapping
- ‘Which one is Noni Fruit?’ - apple, banana, pear known, so it must be the unknown
- Fast Mapping uses shortcut:
- Hippocampus less involved than in Explicit Encoding
- no sleep necessary to induce consolidation of new memory – directly incorporated in place where semantic
knowledge is
- Fast Mapping is artificially constructed procedure (question vs. imperative)
- Long-Term effects of Fast Mapping are weak
- kids are better in this because at age of two hippocampus not fully grown

Exemplar generalization:
- e.g. know immediately what it is: Swexit, Spaxit, Bremain
- Gist extraction: people are able to generalize features from a few (sometimes single) examples
- Generalization is most effective for example-specific features not for general information

- e.g.:

- almost as good in naming the object, when only objects where seen from the same category, as if seen specific item

- Learning via retrieval (abfragen)

- Reconsolidation hypothesis: Information becomes unstable and formable (and consolidates afterwards) if reactivated
(tested)

- e.g. reading and rereading a text or reading and retrieving (abfragen) :

after five minutes first option better. The longer the time after, the better second version
- exposure therapies (e.g. with spiders) are based on retrieval paradigm

- learning from failures

- “problem-solving – instructions”-paradigm: remembering more and understand concept if faced with problem first
- e.g. which bball-player would you chose. Pupils came up with formula (afterwards correct formula presented)
- possible explanations:
- hypercorrection-effect: the bigger the defeat, the bigger the possibility to learn
- reconsolidation of information: false reactivation leads to adjustment of information
- prediction-error: updating mental models if false
- e.g. “hotwashing” navy seals (speak out what went wrong)
- e.g. “purpose training” for soccer players (easy to improve because it is clear where you want to go)

Knowledge
How is semantic memory organized?

Concept: mental representation used for a variety of cognitive functions

Categorization: process by which things are placed into groups called categories (- function of concepts)
- Categories help understand individual cases not previously encountered
- Categories are Pointers to knowledge:
- categories provide a wealth of general information about an item
- allow us to identify the special characteristics of a particular item

Definitional Approach to Categorization

- category membership, when object meets definition of category
- doesn’t work well
- not all members of everyday categories have same defining features (e.g. chair)
- family resemblance (Familienähnlichkeit):
- things in category resemble one another in number of ways

Prototype Approach
- abstract representation of the ‘typical’ member of a category
- characteristic features that describe what members of a concept are like
- an average of category members encountered in the past
- contains most salient features
- true for most instances of that category

Antonio Torralba’s “Average Images”:

- if the pictures being averaged contain commonalities, a pattern emerges revealing the regularities existing in the
intensity patterns across all the images

High-prototypicality:
- category member closely resembles category prototype
- typical member – e.g. for category “bird”: robin
Low-prototypicality:
- category member does not closely resemble category prototype
- e.g. for category “bird”: penguin

Typicality effect:
- prototypical objects are processed preferentially
- highly prototypical objects judged more rapidly
- Sentence verification technique:
- prototypical objects named more rapidly
- e.g. “an Apple is a fruit” shorter RT than “a pomegranate is a fruit”
- problems may be: apple more common than pomegranate (maybe never seen) and word is longer

Exemplar Approach:
- concept is represented by multiple examples (rather than a single prototype)
- examples are actual category members (not abstract averages)
- to categorize, compare new item to stored examples
- similar to Prototype Approach:
- Representing category is not defining it
- different to Prototype Approach:
- representation is not abstract
- descriptions of specific examples
- the more similar a specific exemplar is to a known category member, the faster it will be categorized
- explains typicality effect
- easily takes into account atypical cases
- easily deals with variable categories (like games)

Which Approach works better?

Exemplar:
- deals better with atypical cases
- doesn’t discard information that might be useful later
- e.g. atypical birds can be represented as exemplars rather than getting lost in an average
- deals better with atypical categories like games
Prototype:
- better for large categories
→ Prototype works better when you go for RT, Exemplar better when going for correct answer
→ When initially learning a category, averaging exemplars into Prototypes seems to be useful
Later, Exemplar information becomes stronger and more useful (taking exceptions into account later)

Hierarchical Organization & Semantic Networks

Hierarchical Organization
- there are different levels of categorization from “general” (furniture) to “specific” (kitchen table)

- 3 levels of categories
- superordinate (übergeordnet) / global (e.g. furniture) 3
- basic (e.g. table) 9
- subordinate (untergeordnet) / specific (e.g. kitchen) 10.3

- number: list as many common features as you can that fit all or most objects in category
- from basic to global you loose a lot of information while you only gain a little from basic to specific
- Rosch proposed that Basic level is “psychologically special:
- almost exclusively basic-level names in free-naming tasks
- quicker to identify basic-level category member as member of a category
- children learn basic-level concepts sooner than other levels
- basic-level is much more common in adult discourse than superordinate categories
- different cultures tend to use same basic-level categories at least for living things

- to fully understand how people categorize objects, one must consider:

- properties of objects
- learning and experience of perceivers
- the more expert knowledge, the more specific categories are used

Semantic Networks
- concepts are arranged in networks that represent the way concepts are organized in mind
- Collins and Quillian (1969) Hierarchical network
- Node = category/concept
- concepts are linked
- model for how concepts and properties are associated in mind

living thing
/ ISA \ can
plant grow
/ \
etc…

Cognitive Economy
- shared properties are only stored at higher level nodes
 - exceptions are stored at lower level nodes
- inheritance (Vererbung) → you don’t have to store all information at child node

Sentence verification task (Collins & Quillian 1969)

- task: YES/NO responses to statements about concepts
- experiment with 6 kinds of sentences
- Category sentences
- 0-level-jump: A canary is a canary
- 1-level-jump: A canary is a bird
- 2-level-jump: A canary is an animal
- Feature sentences
- 0-level-jump: A canary can sing
- 1-level-jump: A canary can fly
- 2-level-jump: A canary can breath
- time needed to retrieve information about a concept depends on distance in network
→ RT higher for nodes which are further away

Spreading Activation
- activation is the arousal (Erregung) level of a node
- when a node is activated, activity spreads out along all connected links
- concepts that receive activation are primed and more easily accessed from memory
- e.g. knowledge about canaries, ostrichs may be coactivated by “robin”

Lexical Decision Task

- task: decide as quickly as possible if written stimuli is a word or not
- Myer and Schvaneveldt (1971):
- pairs of words
- “yes”, if both are words, “no” if not
- some pairs closely associated
- RT faster for those pairs → spreading activation

Criticism of Collins and Quillian

- can’t explain typicality effects (e.g. canary, ostrich just one node away, but faster RT for more typical canary)
- Cognitive economy?: e.g. sentence-verification results for model: “pig is mammal” slower than “pig is animal”

Modifications by Collins & Loftus (1975)

- shorter links to connect closely related concepts
- longer ones for less closely related concepts
- no hierarchical structure; based on person’s experience (graph is no tree anymore)

Assessment (Beurteilung) of Semantic Networks

- predictive and explanatory of some results, but not all
- generated multiple experiments
- lack of falsifiability
- no rules for determining link length or how long activation will spread
- therefore, there is no experiment that would “prove it wrong”
- circular reasoning
- if not falsifiable, experiments to prove it is true are also hard to find

Properties of a good Psychological Theory

- Explanatory Power: Behaviour A occurred because…
- Predictive Power: Theory can predict results of a particular experiment
- Falsifiability: should be possible to design experiment that could potentially prove theory wrong
- Generation of Experiments: Good theories stimulate new research to test theory

Connectionist Approach
Neuron-like units
- input units: activated by stimulation of the environment
- hidden units: receive input from input units
- output units: receive input from hidden units

Parallel Distributed Processing (PDP), McClelland & Rummelhart (1986)

- knowledge represented in distributed activity of many units (lots of
information processed at one time)

- weights:
- determine at each connection how strongly an incoming signal will
activate next unit
- correspond (entspricht) to what happens in the synapse that transmits
signals from one neuron to another
- high weights = strong tendency to excite next unit
- low weights = less excitation
- negative weights = decrease excitation or inhibit activation of receiving
unit
(- e.g. if one thinks of a canary, a lot of other cells activated at same time)

- activation of units depends on 2 things:

- signal that originates in the input units
 - the connection weights throughout the network

Difference in coding concepts between PDP and Hierarchical Network

Hierarchical network
- stores information at network’s nodes
Connectionist network
- indicates (anzeigen) properties by activity in property units
- many more activated units per representation
- more complex
- more like what brain actually does (cf. “where are memories located)

Problem-Solving and Reasoning

Some questions to consider
- what makes problems hard?
- is there anything special about problems that seem to be solved in flash of insight?
- how can analogies be used to help solve problems
- how do experts in a field solve problems differently to nonexperts?

What is a Problem?
- Obstacle between present state and goal
- not immediately obvious how to get around this obstacle
- well-defined problems: correct answer, certain procedures will lead to solution
e.g. 13 x 27 = ?
- ill-defined problems: path to solution unclear, not one “correct” answer
e.g. “What shall I study?”

Gestalt Approach
- Representing a problem in mind
- Restructuring: changes the problem’s representation
- Kohler’s circle-problem:

Insight in Problem-Solving
- insight: sudden realization of a problem’s solution
- insight-problem: triangle problem, chain problem
- noninsight: algebra

- triangle problem: chain problem:

- Metcalfe and Wiebe (1987)

- “Warmth” judgement:
- participants asked every 15 seconds how close they feel to solution
- graph: showing how participants judged how close they were to solving at
one minute before actually solving →
- insight-problems solved suddenly
 - non-insight problems solved gradually

Obstacles to problem-solving
Functional Fixedness
- restricting use of an object to its familiar functions
- e.g. candle problem, Duncker (1945)
- task: mounting candle to corkboard so it will burn without dripping
- one group got materials in box, other outside the box
- performance better for group with empty box
→ seeing boxes as containers inhibits using them as supports = functional fixedness
- e.g. Two-String Problem, Maier (1931)
- task: tie together 2 strings; also given chair and pliers
- 60% did not solve
- function of pliers in the way of seeing them as weight
- when one string in motion “accidentally”, almost everybody solved
- solution possible once representation was reconstructed

Information-processing approach
Problem space (Newell & Simon):
- initial state: conditions at the beginning
- operators: rules specify which moves are allowed
- intermediate state(s): sequence of steps towards solving the problem
- goal state: solution of the problem
Means-end analysis: reduce differences between initial and goal state
- Subgoals: create intermediate states closer to goal
- e.g. Hanoi: initial steps show how problem can be broken down into subgoals
(presented as falsy binary tree with crosslinks)

The importance of how a problem is stated

e.g. Acrobat and reverse acrobat
- one small change in wording of problem
- not just analyzing structure of problem space
- both can be solved in same number of steps, but reverse took almost twice as long
- real-world knowledge might interfere with reverse problem

Analogies
Analogy: Process of…
- noticing relationship
- mapping correspondence between source and target
- applying mapping
- often hints must be given to notice connection

Duncker’s Radiation Problem

- high intensity laser destroys other tissue, low intensity laser won’t destroy tumor
- How can tumor be destroyed and other tissue left unaffected?
- Fortress-Story: fortress attacked from many sides, because soldiers would be killed otherwise
→ 3 times more people were able to solve radiation problem

Falsification Principle
e.g. “the Wason Four-Card Problem” (1966)
- determine minimum number of cards to turn over to test:
- if vowel on one side, then even number on other side
-E K 4 7
- to test, one must look for situations that falsify the rule (turn E and 7)
- most participants failed
- Effect of using real-world items in a conditional reasoning Problem
- when stated in everyday terms, correct response greatly increased:
- if alcohol, person must be over 19 (Griggs & Cox (1982))
- cards: beer, soda, 16 y old, 24 y old

How Experts solve Problems

- Experts solve problems in their field faster and with a higher success rate than beginners bc. more knowledge about

- cf. Chess-players

- e.g. grouping together physics problems by novices and experts

- novices group together problems that involve similar objects
- expert group together problemsthat involve similar physics principles

- knowledge is organized so it can be accessed when needed to work on a problem

- novice use surface features (e.g. what objects look like)
- expert use deep structure (e.g. focusing on underlying principles)
- experts spend more time analyzing problem
- experts are not better than novices when given problems outside their field
- experts less likely to be open to new ways of looking at problems

Creative Problem Solving

- Creativity
- innovative thinking
- novel thinking
- new connections between existing ideas
- Divergent Thinking
- open-ended
- large number of potential “solutions”
- Convergent Thinking
- one correct answer
- Design Fixations
- e.g. cup with plastic top and straw
- fixated on what NOT to do as demonstrated by sample
- Fixation inhibits problem-solving
- Creative Cognition
- technique to train people to think creatively
- pre-inventive forms: ideas that precede (vorangehen) creation of finished creative product
- e.g. Finke (90s): parts
- task: make an interesting, unfamiliar object out of… (cylinder, handle, cross)
then interpret as:
furniture, personal item, scientific instrument, transportation, tools, etc…

Language
What is Language?
- system of communication using sounds or symbols
- express feelings, thoughts, ideas and experiences
- language is used for communication, social interactions, convey emotions and stress, maintain social ties
- The speech-chain:
brain (linguistic level) → motor nerves (physiological l.) → vocal muscles (physio. l.) produce sound waves (acoustic l.)
sound waves → ear of speaker and sensory nerves (physiological level) → brain (feedback link, linguistic level)
sound waves → ear of listener and sensory nerves (physiological level) → brain (linguistic level)

Studying Language in Cognitive Psychology

- because language is so pervasive, its study is multidisciplinary
- linguists, psychologists, computer scientists, anthropologists, neuroscientists, biologists, etc. :=
- Psycholinguists discover psychological process by which humans acquire and process language
- Comprehension (Verständnis)
- (Speech production)
- Acquisition

B.F. Skinner “Verbal Behavior” (1957)

- language learned through reinforcement

Noam Chomsky “Syntactic Structures” (1957)

- human language coded in genes
- underlying basis of all language is similar
- children produce sentences that they have never heard and that have never been reinforced
- (language can’t be learned in one kid’s lifetime)
- The Universality of Language
- deaf children invent sign language (strong intrinsic impulsion to do language)
- all cultures have language
- language development similar across cultures (e.g. curves of productive vocabulary)
- languages unique but same: different words, sounds, rules. But all verbs, nouns, negatives, questions, past/present

Both have correct aspects but none is complete on its own: genes can be changed by learning, for example

Language Comprehension (Verständnis)

Perceiving and understanding words
Components of language are not processed in isolation
- Lexicon: all words a person understands
- Phoneme: shortest segment of speech that, if changed, changes the meaning of the word
letters or aggregations like “sch”
- phonemic restoration effect: “fill in” missing phonemes based on context of sentence and portion of words heard
(noise in gaps makes sentences clear again and it feels like the phonemes were there)
- morphemes: smallest unit of language that has meaning or grammatical function
- pizza := /pizza/
- meatball := /meat/ /ball/
- plates := /plate/ /s/

- Speech Segmentation:
- (e.g. didyouseethatgameontvlastnighthowtheywonitatthebuzzer)
- understanding of meaning
- understanding of sound and syntactic rules
- statistical learning (e.g. traffic lights: taking lights of cars into account)
- context is crucial

- Word Superiority Effect (Reicher, 1969)

- context plays important role in interpreting a physical stimulus
- isolated letter such as E or T or word such as KLANE or PLANE is briefly flashed on screen and then immediately
replaced by a mask including E and T
- task: report letter
- independent variable: Letter vs word
- dependend variable: percent correct
- words like PLAN_ used to make sure both E and T make sense
- letter is better perceived within (real!) word than without
- despite increased information, visual clutter (wirrwarr)
→ speaks for beneficial effect of context

- Word Frequency Effect

- respond to high frequent word much faster (cf.: 50 most frequent words in english)
- lexicon for frequent word would look like on first side “this”, while “more” few sides later
- eye movement while reading:
- looking longer on low-frequent words and may skip high-frequent ones

- Lexical Decision Task (LDT)

- string of letters shown
- task: decide as fast as possible, if word or not
- dependent variable: RT

- Priming
- stimulus activates a representation of the stimulus
- one word is priming following ones in LDT
- e.g. you hear auditive stimulus (cat or dog)
- respond more rapidly if you were lexically primed
- word “cat” (prime) coactivates word “dog” (target)

- recent experiment by Fiebach, Gagl (for priming in combination with word frequency effect?):
- same word shown two times or new word
- RT in repetition condition faster
- N400 very much the same for priming words
- N400 much increased (goes deep down) for target, when non-repetition
- N400 reduced for target when repetition

- Context Effects
- attempt to figure out what a sentence means as we read it
e.g. “the cat eats the…” (thinking about a mouse)
e.g. “I scream, you scream, we all scream for ice cream!”

- Lexical Ambiguity
- some words have more than one meaning (Teekesselchen)
- after all meanings of a word have been briefly accessed, context clears up ambiguity
Experiment by Swinney (1979)
- “This government building had been plagued with problems for years. He was not surprised to find spiders, roaches
and bugs in the corner of the room“
- had people listen to paragraph
- when “bug” was heard, he presented a word on the screen
- word either related to “insect” (ant), “hidden listening device” (spy) or unrelated (sky)
- LDT: words “spy” and “ant” were primed by “bug”
→ RT was much quicker than for “sky”
- parallel access to both meanings while “bugs” was heard

- same experiment with delay of 200 ms to “bug”

→ only “ant” was primed
→ shows effect of sentence context to resolve ambiguity

- Understanding Sentences
- Semantics: meaning of words and sentences
- Syntax: rules for combining words into sentences
- event-related potential and brain imaging studies have shown syntax and semantics are associated with different
mechanisms
- Semantic violation:
- potential N400 (negative at 400 ms)
- semantic memory / world knowledge
- integration
- Syntactic violation:
- potential P600 (positive at 600 ms)
- rule based / non-automatical
- reanalysis
- e.g.: Every morning at breakfast the boys would eat …
Every morning at breakfast the boys would plant ...
Every morning at breakfast the boys eat would …

- Parsing: mental grouping of words in a sentence into phrases

- syntactic ambiguity: more than one possible structure, more than one meaning
- Garden path sentences: grammatically correct sentence that starts in such a way a reader’s most likely
interpretation will be incorrect – reader is lured into a path turns out to be a dead end
e.g. “the old man the boat” means that the boat is manned by the old

- Understanding Text and Stories

- coherence: representation of the text in one’s mind so that information from one part of the text can be related to
information in another part of the text
- inference: readers create information during reading not explicitly stated in the text
- Situation Model: mental representation of what a text is about
- represent events as if experiencing the situation
- point of view of protagonist
- e.g. task: decide if picture matches the object mentioned in the sentence
- orientation experiment: “He hammered the nail into the floor” (vertical/horizotal nail)
- shape experiment: “the ranger saw the eagle in the sky” (sitting/flying eagle)
- RT faster, when picture fits sentence
→ supports idea that participants create perceptions that matched the situation as they were reading the sentences

- Culture, Language and Cognition

- Some forms of thought seem more linguistic than others
- Are language and thought inter-dependent?

Sapir-Whorf-Hypothesis (1956):
- language entirely shapes our thoughts, and the way we see the world
- Whorf concluded that thoughts are not free – they are bound by language
- Linguistic Relativism:
- The hypothesis that thinking shapes language AND language shapes thinking
- the structure of a language effects the ways in which its speakers conceptualize their world
- varying perceptions that exist among speakers of different languages are apparent in spatial and color perception
- the validity of linguistic relativism does not seem to be as strong as the Sapir-Whorf Hypothesis once suggested
Language in the Brain
- language production is impaired (beeinträchtigt) by damage to Broca’s Area, in Frontal Lobe
- language comprehension is impaired by damage to Wernicke’s Area, in Temporal Lobe

Broca Aphasia
- impairment of language ability
- develop quickly as a result of head injury or stroke
- loss of ability to produce language
- speech is difficult to initiate (beginnen), non-fluent, labored (mühsam), and halting

Wernicke Aphasia
- impacts language comprehension and the production of meaningful language
- difficulty understanding spoken language but are able to produce sounds, phrases, and word sequences
- while these utterances (Äußerungen) have the same rhythm as normal speech, they are not language because no
information is conveyed

Aphasia in Sign Language

- sign language contrasts cerebral hemispheres
- left hemisphere is dominant for language, but the right is dominant for spatial relations
- Bellugi (1987) found that a damaged left hemisphere leads to an ASL (American Sign Language) version of aphasia

Lateralization
- is a well-known characteristic of the brain; asymmetry
- Cortex is divided into two hemispheres, connected by corpus callosum
- corpus callosum is largest bundle of nerve fibers in brain
- carries messages between the hemispheres
- some brain functions have evolved to be lateralized to one side
- in most people, left hemisphere is language-dominant, which was revealed in split brain-patients
- Sperry (1968) performed surgery to relieve (lindern) epilepsy, severing (trennen) the corpus callosum
- hemispheres were limited, when in isolation
- participants were shown an object in their right or left visual field
- perceived by opposite brain hemisphere
- picture on right side → left hemisphere
- can name object
- could never draw it or find it by touch
- picture on left side → right hemisphere
- can draw it with left hand (and closed eyes) and find object by touching it
- objects touched with left hand (right hemisphere) can’t be named (switch hands: no problem)

Language Acquisition
- Process by which capacity (Fähigkeit) is acquired to perceive (wahrnehmen) and comprehend language, as well as to
produce and use words to communicate
- To use language, range of tools must be acquired, including syntax, phonetics, and an extensive vocabulary

Nature vs. Nurture

- language onset (Beginn?) is correlated with age (critical period)
- even with biological “programming” for language, experience is needed during formative years
- Emphasis (Schwerpunkt) on Nature
- innate biological “device” for understanding principles and organisation common to all languages
- theory of universal grammar (Chomsky)
- Emphasis on Nurture
- greater emphasis on the influence of usage and experience
- argue that adults play an important role in language acquisition by speaking to children

Animal Language
- Experiment by Herb Terrace
- students raised chimp “Noam Chimpsky (Nim)”, teaching it a simple version of ASL
- Nim learned many signs and had rudimentary grammar
- double-blind analysis of videotaped interactions (double-blind: neither participant nor researcher knows about which
participant gets which treatment)
- discoveries:
- Nim almost never created own sentences
- Nim gave simple responses to trainer’s signs, typically for rewards
- no consistent grammar
- signed almost randomly until rewarded
- Nim’s language was basically complex operant conditioning (operant conditioning – like Pavlov)
- Kanzi (monkey) had icons to point on, when wanted to have stuff from experimenter
→ animal language very different to how we learn
- lot of words could be learned
- bad at combining words

What would happen if a child was raised with no linguistic input?

- Victor, wild-boy of Averon
- found 1800 in France wood, naked
- around 11 or 12, had apparently been alone for years
- couldn't speak, producing only guttural noises
- after month of teaching, playing games, socializing, he could read a few words (matching to pictures)
- despite he seemed intelligent, he never learned syntax and he never learned to speak

Decision-Making
- What kind of traps do people get into when making decisions?
- How do emotions Influence decisions?

Decision Tree
- summarizing components of a decision
- branches are different courses (alternatives) of action, choices and strategies
- belief: estimate of the likelihood that particular outcome will occur
- consequences: benefits or losses

status quo, sure thing

risky option

uncertainty

Heuristics

Availability Heuristic, Tversky & Kahnemann, 1973

- events more easily remembered are judged as more probable than those less easily remembered
- e.g.: more english words with “r” as first or third letter? // 70% judged false // “r” is 3 times more often third letter
- e.g.: asthma-deaths are 20 times higher than tornado-deaths // 58% judged death through tornado as more likely

Illusory Correlations
- correlation appears to exist, but doesn’t or is much weaker than assumed
- Stereotypes:
- oversimplified generalization about group or class
- often focuses on the negative
- develop because more attention to behaviors associated with stereotype
→ illusory correlation that reinforces (verstärkt) stereotype is created
- Related to availability heuristic: selective attention to stereotypical behaviors makes them more available

Representativeness Heuristic
- probability that B comes from A can be determined by how well B resembles properties of A
- (how well does B fit into prototype A?)
- e.g. randomly picked male from US population: Robert wears glasses, speaks quietly, reads a lot. Farmer or Librarian?
- base rate for farmers much higher
- use base rates if nothing else available
- use descriptive information if available and disregard base rate information
- (example is not about stereotypes)

Conjunction Rule
- probability of two events cannot be higher than probability of single consistent
- probability of having a car cannot be smaller than of having a red car

Law of large numbers

the larger number of randomly drawn individuals from population, the more representative for entire population

Confirmation Bias
- tendency to selectively look for information that confirms (bestätigt) our hypothesis and overlook information that
argues against
- Lord (1979):
- people in favour of and people against capital punishment read same article
- those in favour found article in favour
- those against found article against

Economic utility theory

- people are rational
- if they have all relevant information, decision will result in maximum expected utility
- utility: refers to desirable outcomes that are in person’s best
interest (e.g. monetary payoff)
- advantages (Vorteile) for utility approach:
- specific procedures to determine “best choice”
- problems for utility approach:
- not necessarily money, people find value in other things
- many decisions involve payoffs that cannot be calculated
- Kahnemann & Tversky (1979) – utility curve
- a loss “hurts” more as twice as much as a gain is valued
- more losses hurt more, but diminishing effect, as losses accumulate
- same for gains

Emotions and Decision-Making

- emotions affect decisions
- expected emotions
- immediate emotions
- people inaccurately predict their emotions
- think they would feel very bad about a loss and just a bit gut about a gain →
- but actually they were a bit more happy after a gain than unhappy about a loss

Framing effect
- decisions depend on how choices are represented
- can highlight one aspect of situation

- opt-in procedure (e.g. active step to organ donor)

- opt-out procedure (organ donor unless request not to be)

- risk-aversion strategy
- used when problem is stated in terms of gains
- risk-taking strategy
- when problem is stated in terms of losses
- e.g. Program A saves 200 people (out of 600)
Program B: 1/3 chance 600 saved, 2/3 chance None saved
- stated in terms of gains → risk-aversion → Program A chosen
Program C: 400 die (out of 600)
Program D: 1/3 chance None die, 2/3 chance 600 die
- stated in terms of losses → risk-taking → Program D chosen
- A=C and B=D