Cog Psych

Question 1

What is cognition?

Answer 1

• An activity of the mind
• Fundamentally involves acquiring and using knowledge
  ○ The mind is a system that creates representations of the world so that we can act within it to achieve our goals

Question 2

What are the mental process involved in cognition?

Answer 2

§ Perception
			§ Memory
			§ Attention
			§ Decision-making
			§ Reasoning
			§ Problem solving
			§ Imagining
			§ Planning
			§ Executing actions

Question 3

When is a system cognitive?

Answer 3

• It must coordinate its behaviours with environmental features that are not always present
  □ Cognitive agents interact with environment in ways that require them to refer to experiences that aren’t necessarily present right now
  □ Rely on past experiences
• In such cases, there is something else that stands in for the non-present thing
  □ Mental representation
  - Parts of a general representational system that
  allows standing in to occur systematically in
  relation to other related representational
  states

Question 4

What does it mean for cognisers to have agency?

Answer 4

○ Able to explore their world in some way
§ Sense and act on their environment to
□ Detect and effect changes
□ Gain information
○ Construct mental models to represent the causal structure of their environment
§ Mental model = mental representation that allows
you to represent a number of concepts in relation
to one another in a model of the world
○ Adapt their mental models in response to feedback from their behaviour
§ Constantly evolving and changing models
○ Use mental models to guide future behaviour
○ Form inferences to make sense of experience

Question 5

What is the Turing test/imitation game?

Answer 5

○ Two people in rooms separate from the people playing the game
○ Pass notes under the doors: one person is male, one is female - person playing game needs to work out which is which based on answers they give on the paper
§ Both are trying to convinve the person that they
are female
○ Replaced the one of the people with a computer: had to guess whether the responses given were from a human or a computer
○ Computer passed the Turing test if the person guessing couldn’t distinguish which was human and which was computer (better than chance)
○ Given a particular linguistic input, if the machine can provide appropriate linguistic output than it could potentially pass the Turing test
○ Tests the idea that human cognition is somehting like computation
○ Psychology was evolving alongside with the computer

Question 6

Describe the computational account for cognition

Answer 6

• The digital computer as a metaphor for thinking, reasoning, problem-solving:
○ The mind is the software
§ Runs programs based on algorithms
§ Sequential processing: If…Then…
○ The brain is biological hardware
• Thinking is the mental manipulation of symbols according to syntactic rules
○ Symbols represent our knowledge of things
and events (concepts)
○ Syntactic rules enable us to express the
relations between symbols
• Thinking is represented in a ‘metalese’
• Natural languages translate mentalese into a publicly expressed format
• Propositional representations are the basic units of mentalese
• Good for explaining some of the reasoning methods that humans engage in
○ Eg puzzles

Question 7

What is mentalese?

Answer 7

○ A language of thought

○ Cognition takes place in a mental language

Question 8

According to the computational account, are computers cognisers?

Answer 8

• This approach does permit AI as cognisers, with the brain being simply one physical system capable of running a program

Question 9

What are propositional representations?

Answer 9

• A symbolic code to express the meaning of concepts, and the relationships between concepts
• Images can be expressed in the natural language (ie a description)
○ Eg ‘the cat is on the table’
• Or the underlying propositional representation of the relationship between concepts can be expressed as a propostion
○ Eg ON (CAT, TABLE)
§ Where ‘ON’ is the predicate stating the
relationship between the two arguments
(‘CAT’ and ‘TABLE’)
□ Arguments are the roles within the
concept
§ P (x, y)
• How we can represent the core meaning of any particular situation
• The same abstract porpositional frame/schema can express many different surface forms
○ Gave (agent, object, recipient)
§ John gave Mary the book
□ Gave (John, book, Mary)
§ The book was given to Mary by John
□ Gave (John, book, Mary)
• Propositions can be combines to represent more complex relationships

Question 10

What is the situation model?

Answer 10

• Situation model = mental model
• ‘The turtle sat on the log. The fish swam underneath’
○ What did the fish swim underneath?
○ We infer that the fish swam underneath the log
○ The situation model of past experiences allows us to make inferences
• Propositions allow inferences/reasoning
○ On (Turtle, Log)
○ Swim (fish)
§ Under (Log, fish)
§ Below (Turtle, fish)
• Inferences rely on connecting the input with prior knowledge
• This is where a computer might fail the Turing test: how does it know what inferences to make?

Question 11

What is a semantic network?

Answer 11

• Semantic network model = the set of concepts and ideas that inhabit a human mind
• Broad concepts can be defined by other concepts
○ Understanding of the meaning of the concept is in relation to the other kinds of entities in the world that I have experienced in relation to that concept
• Mind is built off a set of symbols that stand in for the things they refer to in the world

Question 12

Describe how Shepard and Metzler’s mental rotation study demonstrated a Dynamic account for cognition

Answer 12

○ Challenges the idea that linguistic/language-like mentalese underpins all cognition
○ Wanted to think about intuition
○ Mental representations can be analogous to sensory experiences of the world
○ In the experiment, the response time for correct responses increased as the degree of angle rotation of the object increased from the original image
- Implies that at least some of our mental representations are carried out using analogue representations (visual imagery in this case) rather than abstract symbols (eg propositional representations)
- Mental images are analogous to what they represent
□ More direct representations of the things we’re thinking about
- We manipulate mental images in our minds in a manner which is analogous to the way in which we might physically manipulate a real object
○ Dynamic cognition because the timing and the continuous nature of the rotation suggests that unlike the stage-like problem solving of a linguistically-based computer algorithm, we have the smooth handling of an object in space in real time
§ Time matters
• Analogue representations
○ Analogous to the things they represent
○ Mental imagery is key
○ Direct opposition to symbolic linguistic representations proposed in computational account

Question 13

Describe how Spivey and Dale’s mouse-tracking study demonstrated a Dynamic account for cognition

Answer 13

○ Interested in shaking up the idea that cognition occurs in discrete symbolic representations
○ Study:
§ Participants looking at two images
§ Asked to move computer mouse from a starting
point towards the picture being named
§ Used the trajectory of the mouse-movements to
show the unfolding of cognition over time
§ Two conditions: cohort (conflict), and control (no
conflict)
§ The conflict trials used images with similar names
(eg candle and candy)
§ Found that the trajectory of the control condition
was much more direct right from the start
§ In the cohort condition, the trajectory tracked more towards the centre as the word unfolded, and took a turn when the word was completed, thus the distinguishment between the two names was made clear
□ Emphasised the role of cognition unfolding continuously and in real time, and of it being something that is not just a mental process, but also the embodiment of the response dynamically alongside the cognition

Question 14

What are the Situated and Embidied accounts of cognition?

Answer 14

• Situation and environment can support cognition

* Cognition does not only occur in your head, but also in continuous interaction with the situation

Question 15

How did Brooks use robots to support the Situated and Embodied accounts of cogntion?

Answer 15

○ Began AI projects with simple robotic systems designed to navigate simple enviornments
○ ‘Mobots’ modelled off insects
○ Sense the world around them and adapt to changes in their environments
○ Foundations of cognition
§ Other cognitions are built of theses basic sensory
responses and adaption
§ Sensation and perception are at the core of basic
categorisations of whether the robot will
approaches or avoids situations
○ Argument that we need both types of representations:
§ It’s not that minds work with just mental imagery of
language or sensory perceptual types of
embodied representations
§ Minds use all of these types of representation and
crucially build on one another:
□ Top level where symbols are depends on the
learning that occurs at the lower levels to
build meaning into the symbols
□ Building cognition from the bottom up

Question 16

Explain the use of Leonardo the social robot

Answer 16

○ Emphasis is not on the symbolic, linguistic level - he doesn’t use/understand objects
○ Can recognise objects, facial expressions, and vocal tone
○ Represents a non-linguistic child /pet that can pick up on social and emotional cues from the environment as a way of learning about objects and their environments
○ Can do shared attention
○ Has mirror neurons: can mimic facial expressions shown to him
§ A way of telling the computational system the type of emotion attached to the objects being introduced to it
○ Has object appraisal mechanism - knows when objects are new/unfamiliar
○ When things are new - a mild anxiety response is triggered
§ Leo will show questioning/concerned facial
expressions
§ Promotes humans to explain the situation to Leo in
response to his reaction
§ Leo detects vocal tone and facial expressions to
help assess the situation/object
○ Attentional system in robot monitors the human facial expressions and focal point and looks between human and object - shared attention
○ Empathic mechanism mirrors human facial expressions to stimulate emotion in itself
§ When Leo encounters the object again, its
response will mirror that of the human’s
○ Change in emotional state triggers long-term memory response for object, tagged with socially referenced emotional informaiton
• Shows that the embodiment, social interactions, and the learning from these interactions allow us to make basic categorisations about emotional responses towards objects
○ Foundation of building meaning and concepts

Question 17

What is the Cartesian approach to cognition, and what are the contentions?

Answer 17

• Computational approach is a cartesian approach
• Mind is disembodied and operates on abstract plane of
  symbolic representations without embodiment,
  emotions, or interactions
  -Describes some of our logical reasoning capacity, but
  does not explain how they get there and how the
  symbols get their meaning
  -Instead of saying ‘I think, therefore I am’, should say ‘I
  feel/sense, therefore I think’
  -Abstract conceptual knowledge (eg love, peace) must be grounded in our perceptions and interactions with the world
  -ignores how symbols acquire their meaning

Question 18

What is the hierarchy of mental representations?

Answer 18

• Top = symbolic representations, schemas, propositions, narrative
• Middle = visuo-spatial representations, mental imagery
• Bottom = direct sensorimotor representations of current experience

Question 19

Describe the perceptual cycle

Answer 19

• In any given moment, our experience is a product of integrating the perceptual present and the cognitive past
○ Implies an active, embodied agent embedded in the physical and socio-cultural world
○ The sequential-cyclical process belies a deeper embedding and inter-dependence of brain, body, world, and mind
○ Doesn’t quite capture the three-dimensionality of the deep embeddedness, but gets close

Question 20

What is a spoken word?

Answer 20

an association between a co-occurring sequence of speech sounds and an entity, event or idea to which it refers

Question 21

Describe the study on Vervet monkey communication

Answer 21

○ Vervet monkeys have three distinct calls associated with the threetypes of predators
§ Eagle
§ Leopard
§ Snake
○ Three distinct escape behaviours associated with the three calls
○ Infant vervets initially generalise calls to similar stimuli
§ They gradually learn to discriminate which entities each sounds indicate more specifically
○ Are the calls the same as words in a language?

Question 22

What is language?

Answer 22

○ A mode of communication based on symbolic reference
○ Involves combinatorial rules that comprise a system for representing logical relationships among these symbols
○ Two core components of language (forms of signaling are different from forms of language):
§ Word meaning/semantics
□ Symbolic reference
§ Syntax
□ Rules for combining words into sentences

Question 23

When does communication occur?

Answer 23

• Communication occurs not only between members of the same species, but also thinking to yourself is a means of communication in itself
  
  • The structure of thought, and the structure of language have some overlaps

Question 24

What is Reference?

Answer 24

○ There is some signal in the environment that causes me to represent to myself another idea to which it bears some relationship
○ Reference = the means by which one thing (a sign) brings to mind another thing (the signified)

Question 25

What are Pierce’s three modes of reference?

Answer 25

□ Iconic
□ Indexical
□ Symbolic
§ The three modes of reference are defined by the nature of the relationship between the sign and that which is signified

Question 26

Describe the iconic mode of reference

Answer 26

○ Based on physical similarity between the sign and what it signifies
§ A portrait or landscape paintnig relies on the physical similarity between the painting and that which is represented
§ Pantomime relies on similarity between actions and object/situation being portrayed
§ Onomatopoeia in spoken language relies on an iconic resemblance between the sound of the word and its referent
○ Iconic reference forms the basis for basic perceptual recognition processes
§ Re-cognition
○ And for perceptual forms of mental representation (mental imagery)
§ Re-presentation
○ Iconicity underpins how we ‘re-cognise’ similarity between the current input and previous experience

Question 27

Describe the indexical mode of reference

Answer 27

○ Whenever the form of reference is some kind of co-ocurrence
○ Based on a physical or temporal co-occurrence between a sign and what it signifies
§ Alarm could be an indicator of a predator through temporal co-occurrence
§ Smoke indicates fire - physical co-occurrence
§ Word used in connection with an object or event - learning word meanings
§ Co-occurrent learning
○ Indexical reference relies on iconic reference
§ Indexical reference works by recognising some recurring relationhsip between two iconic representations
§ Eg rat in the Skinner box - rat learns to associate the tone with the shock
§ Requires forming an iconinc representation of the tone and the shock stimulus, and then representing the correlation between them
§ The tone indexes the shock
○ Essentially, indexical reference is relating two iconic references

Question 28

Describe the symbolic mode of reference

Answer 28

○ Relies on the capacity for both iconic and indexical modes of reference
○ We refer symbolically when a word’s meaning is understood in terms of its relationship with other words as they occur in sentences
○ Require us to be able to recognise the correlations between indices
○ Symbols are grounded in iconic and indexical representations
○ Eg how does cat relate to dog? - finding the connection between the two indexical modes of reference for each individual animal
○ Relationship between words

Question 29

Describe symbolic learning

Answer 29

§ To learn symbols we begin by recognising iconic forms and symbol-object correlations
□ Iconic relationships between one instance of a word and another become important to establish reliable word recognition
§ Indexical relationships are crucial initially to establish the link between a word and its referent
□ The occurrence of the word cat is initially associated with the presence of an actual cat
§ Symbolic learning is hard because what determine the meaning of the symbol is not the probability of co-occurrence with its referent in time or space
□ Also difficult because words with simlar meanings are often used alternatively
□ Words with different meanings tend to be used together in a sentence
□ The frequency with which specific word combinations are repeated is extremely low
□ In basic learning theory, such reliable relationships would eb extinguished
§ Instead, symbolic meaning is a complex function of the relationship that a symbol has to other symbols within a symbolic system
□ Words are defined by their relationship to other words
□ Higher-order correlations between words within a grammar begin to matter more than the co-occurrence of a word and its referent in time and space
□ Eg. Learning ‘dog’, ‘cat’, ‘animal’ - learning what a dog is also relies on learning that a cat is not a dog

Question 30

What is the relationship between the three modes of reference?

Answer 30

• essentially: indexical references are the relationships between iconic references, and symbolic references are the relationships between indexical relationships
-Hierarchy: symbolic at top, indexical in middle, iconic at bottom

Question 31

What differentiates human infants learning from chimpanzees learning?

Answer 31

• Something that differentiates human infants learning language from chimpanzees learning everythign they learn in a natural environment is that human infants have a predisposition towards paying attention to the symbolic level whereas chimpanzees find it difficult to shift their attention to the symbolic attention and away from the operant rewards

Question 32

Describe Terrace’s study on raising chimps and what the finding imply for whether apes can learn symbolically

Answer 32

§ Set out to look at what would happen if you raised a chimp with humans and tried to teach it a human language (sign language)
§ Nim (the chimp) could use sign language, but did he understand what the signs meant symbolically - is he using them indexically or symbolically?
§ Conclusion was that although Nim could communicate words through sign language, the underlying meaning was not what he communicated
□ Did not use them symbolically but instead indexically to get what he wanted
§ Decided that this meant chimps could not learn symbolically

Question 33

Describe Savage-Rumbaugh’s contention to Terrace’s chimp study, and her improvement on his study

Answer 33

§ Felt that the problem with experiments like Terrace’s was that they were hell-bent on the chimp using an entire sentence - their views were that words alone were not sufficient, and the syntax of creating a sentence was the most important thing to demonstrate language
§ Savage-Rumbaugh felt that this pushed the animals to syntax before developing the iconic and indexical relationships found symbolic references
§ Thuoght that people needed to look at the cognitive competencies underlying symbolic processing, rather than looking for superficial similarities between chimps and humans
§ Her take on symbolic learning
□ Used lexigrams rather than symbol for the chimps to communicate with
□ Studied two young male chimps
□ Wanted to find out whether a chimp knows that a symbol can stand for an absent object
® Focus on comprehension rather than production
® And symbolic reference rather than syntax
□ Lexigrams were shuffled around, so they were actively looking for the symbol rather than the position

Question 34

What is a lexicon?

Answer 34

○ Mental dictionary

○ The store of knowledge for the words we know and their meanings

Question 35

What is phonology?

Answer 35

The auditory/spoken perceptual form of words

Question 36

What is orthography?

Answer 36

The visual/written perceptual form of words

Question 37

What do the perceptual form of words map onto?

Answer 37

lexical representations (lemmas) and concepts (semantics)

Question 38

What is the main issue with spoken word recognition?

Answer 38

• When you’re trying to comprehend speech - confronted with rapid sequence of sound produced sequentially
• Speech has a critical temporal nature to it - we have to process words as they unfold as sound waves in real time
• There are no gaps between spoken words
• Need to try and identify individual words within a continuous stream of speech
• On average, spoken words take approximately 500 milliseconds to unfold
○ Presents problems with working memory capacity - losing access to memory of words preceding it
○ Need to predict word ahead of it finishing
• Word recognition would be simple if lexical selection (the process of recognising words) began only after a word finished
○ Would not allow us enough time to process the meaning of speech
○ Important that the cognitive models of perception of speech are realisitc in terms of not being sequential symbolic processing (computation account), and intead are dynamic models that allow the brain to form multiple hypotheses about what it’s hearing ahead of time

Question 39

What is the solution to the problem of spoken word recognition?

Answer 39

• speech is evaluated and reevaluated continuously against numerous potential options (lexical competitors): hypotheses
  ○ Influenced by context
  • This phenomenon is called the parallel activation of a cohort of multiple word forms and their semantic associations
  ○ When you hear the beginning of a word, this sets off a chain of events where a number of different potential matches of that word become active in your mind in parallel
  § Parallel activation
  ○ Cohort = group of potential words

Question 40

Explain the cross-modal cohort priming effects as evidence for parallel activation

Answer 40

§ Experiment
□ Hear a ‘prime’: the stimulus that is given to you
ahead of another task you need to do (eg the
sound ‘d’)
□ While listening to the prime, shown words and you
need to identify whether it is a real word or not
□ Cross-modal because it is auditory to visual
processing
□ The lexical decision (priming effect) will be faster if the word appearing begins with the same letter as the sound is making
® Indicates priming of the cohort of words starting with that letter: parallel activation

Question 41

Explain the cross-modal semantic priming of cohort as evidence for parallel activation

Answer 41

§ Semantic priming occurs for words related in meaning to the phonological cohort
§ Partial word recognition results in rapid activation of both the phonological cohort, and words related in meaning
§ Why is this useful?
§ Experiment
□ Start a trial by hearing a fraction of the beginning of a word, and your brain begins processing a variety of words that start with that fraction
□ Shown words that are related to words that could start with that fraction
□ Eg could hear ‘Capt’ (could be captain, captive), and shown words that relate to those (eg ship, guard)
□ Lexical decision for words that are related in meaning to the beginning fraction is faster than unrelated words

Question 42

What is the trace model of spoken word recognition?

Answer 42

• Computational model - not in classical sense as the original computational account
• McClelland and Elman study
○ Aimed to model recognition processes in terms of a connectionist network
○ Has interactive activation framework
○ Interactive activation connectionist model of speech recognition
○ Model simulates the kinds of cohort priming effects we see in the previous studies, and provides a mechanism for ways the best lexical match to the input is found
○ Our minds have formed representations for words that are familiar, so when we hear words spoken in real time, there is a matching process between the input and a stored representation in lexical memory
○ Three layers within the network
§ Bottom up recognition -
§ Hear the input
§ As we hear it, the first info minds are getting when processing the sound is to do with the structure of the sound itself
□ Features: eg voicing (whether a sound is voiced or not: b is voiced, t is not)
□ Earliest stage of processing the physical input
□ Each feature has a node
□ When a node is activated (so with the letter b , the voicing node is activated), it sends an activation signal to the next level up (phonemes), that it can begin to recognise
§ Next level up = phonemes
□ Instead of having parts of speech sounds, we have recognisable speech sounds like specific letter sounds
□ Phonemes that build activation send their activation up to the lexical/word level
§ Connectionist network: looking at connections between nodes at each level and in the system that builds up to help us recognise individual words that are represented at the lexical level
○ Each node has a resting level of activation
§ Reflects its frequency of occurrence within the system
○ Connections are trained with a training set: words in the English language
§ Model is exposed to each word proportionally to the relative frequency as they occur in the language
○ Every node connects to every other node
○ As activation accumulates in multiple nodes of at the lexical level, the nodes compete via lateral ihibition
§ The mechanism within the trace mode that allows the best match to the input to be selected
§ To look at that need to look at diagram and look at arrows represented across lexical layer
§ Arrows that appear between lexical representations compete laterally - not excitatory, but inhibitory
□ Suppress activation in another node
○ The winner is the lexical node with the strongest activation once all excitatory and inhibitory inputs are accounted for - most amount of evidence
○ Lateral inhibition at lexical level allows target selection

Question 43

What are the four theoretical viewpoints on bilingual word recognition models?

Answer 43

○ A: Language-selective access, independent lexica
§ Hypothesis that bilingual minds are both selective (ie input will only activate one language) when the input is able to be identified, and activates only that language in the lexicon level
§ Lexica of the two languages are independent from each other
○ B: Selective access, integrated lexica
§ Input activates only the language the input is in (the input is processed in the language it is presented in)
§ Lexicons are integrated - there is a single lexicon that covers both languages (ie the lexical level has both language options)
○ C: Language non-selective access, independent lexica
§ Input could be processed in either language (eg for similar sounding words across languages - both languages will be activated)
§ Lexica are independent
○ D: Non-selective access, integrated lexicons
§ Both languages are all housed together in the one representational place
§ Single lexicon for all the words we know in both languages
§ Whenever speech is being processed, there will be non-selective access every time regardless of context

Question 44

Describe Marian and Spivey’s study on bilingual word word recognition/processing

Answer 44

○ Tracked eye trajectories
○ German-English bilinguals doing a task in english
○ Participants are asked to select from 4 options which picture represents the word that was said to them
○ Need to click on the pic
○ Two conditions
§ In one, the word they hear is the beginning of an english word that matches one of the images, but also shares its sounds with a german word that means something different
□ Looking at whether bilinguals have language-non selective processing and a shared lexicon
□ Evidence for this show that even when operating in English, their eyes are attracted to both the images that match the English and German words
□ Shows that the automatic processes activate both English cohort, and German cohort words
□ Language is being processed bottom-up (from input up), regardless of top-down expectations
§ In the other condition, the beginning of the word only matches the English word for one of the images
○ Shows evidence that bilinguals process language in the way described by D: language-non-selective access, integrated lexica

Question 45

Describe Dijkstra and Van Heuven’s study demonstrating the bilingual interactive model

Answer 45

○ Did a similar study are Marian and Spivey, but with written language
○ Showed similar effects:
○ In the written context, rather than cohort effects, they’re neighbourhood effects
○ Even when it isn’t relevant to the task, one language’s input will activate neighbours of the other language’s
○ words that are related to it

Question 46

Describe the control processes in bilingual language recognition and production

Answer 46

• Because of the language non-specific access and integrated lexica, bilinguals need to control
○ The intention to speak the target language
○ Selection of words in the target language
○ Inhibition of words in the non-target language
○ Monitoring for intrusions from the non-target language
○ Language disengagement and engagement (ie switching between languages)
• Theory is that since bilinguals need to use these control processes to monitor language, there would be a carry-over effect into other tasks requiring similar levels of control - bilingual would have an advantage of the monolingual in the executive control processes involved in controlling intention to do one thing or another, particularly in the context of competing information

Question 47

Describe Green’s inhibitory control model of bilingual language production

Answer 47

○ Bilingual lexico-semantic system (entire set of words that a bilingual knows in both languages, and semantic system - conceptual system the words map onto
§ Shared lexicon of words for both languages that map onto a single conceptual system - concepts will have 2 word forms that map to them
○ I = input
○ O = output
○ G = current goals as a cogniser
§ Domain general/central systems of the mind that allow us to control particular domains
□ Language domain in this case
○ The upper level (G, Conceptualiser, SAS), are the central executive
○ Conceptualiser - allowing us to conceptualise the goals and think about them in terms of representations of imagery or words
○ SAS = Supervisory Attentional System
§ Central executive system - allows us to control behaviour to meet goals
§ Exerts inhibitory control to enable selection of schemas relevant to the goal
○ SAS activates the Language Task Schemas
§ Allow us to select intended language by inhibiting the activation of task-irrelevant tags
§ Representations of the kinds of things you can do with language relevant to the task at hand
○ If you have an intention to speak about a particular thing in a particular language, the concepts will automatically activate the words in the lexicon that relate to it
§ As soon as you start thinking about what you want to talk about, there is an automatic activation of the lexicons in both languages
§ Therefore, lexico-semantic system has two ways of input: coming from outside world, or from one’s own intentions
○ Words within the system are tagged as L1 and L2 according to language
○ SAS is slower to respond - controls which words become dominant
Inhibition of SAS is reactive to the automatic process occurring in the lexico-semantic system