Selective attention

Question 1

dividing attention between voices - what is the cocktail party problem?

Answer 1

Cannot understand/remember contents of 2 concurrent spoken messages (Cherry, 1950)

Best we can do is to alternate between attending selectively to speakers

People with headphones, where’s the bottleneck? – where do we get stuck?

Question 2

focused attention to one of 2 simultaneous speech messages

Answer 2

2 messages through headphones - ‘shadowing’ (repeating aloud) on of messages is:

Successful if messages differ in physical properties (location, voice, amplitude)

Not successful if differ only in semantic content (e.g. novel v recipe/2 categories of word)

What kind of changes do Ps notice in unattended message?

Physical changes (location, voice, amplitude, gross phonetics – e.g. English  Czech)

Not semantic changes (e.g. from meaningful to meaningless – such as English  Czech-pronounced-like-English, words to pseudowords)

Word repeated 35 times in unattended message not remembered better (in later unexpected recognition test) than word heard once (Moray)

Question 3

does attentional selection precede lexical identification and access to meaning?

Answer 3

Seems that unattended words ‘filtered out’ early, after analysis of physical attributes, before access to identity/meaning

Though aware of unattended speech as sounds with pitch, loudness and phonetic characteristics (in background), don’t seem to process identity/meaning

If required to extract identity/meaning from 2 sources, P has to switch attention filter between them – slow and effortful process:

Broadbent’s (1958) dichotic split-span exp:

Left ear: 3, 7, 6

Right ear: 9, 4, 2

Easier to report as 3 7 6 9 4 2 (one switch)

Than 3 9 7 4 6 2 – switch for each digit

Question 4

what is Broadbent’s (1958) ‘filter’ model?

Answer 4

Sensory features of all speech sources processed in parallel and stored briefly in sensory memory (for auditory input: ‘echoic memory’/’pre-categorical acoustic storage’)

Selective filter directed only to once source at time

Filter early in processing so that only info that passes through filter achieves:

Recognition (of words, objects, faces etc)

Activation of meaning

Representation in memory (WM and LTM)

Control of voluntary action

Access to conscious awareness

2 additional assumptions of Broadbent’s account:

Filter is all-or-none

Filter is obligatory ‘structural’ bottleneck

Question 5

is filtering all-or-none?

Answer 5

Examples of partial ‘breakthrough’ of meaning of unattended speech in shadowing exps:

Own name often noticed in unattended speech (Moray, 1959)

Interpretation of lexically ambiguous words in attended message influenced by meaning of words in unattended message (Lackner and Garrett, 1972)

Attended: The spy put out the torch as out signal to attack

Unattended: Extinguished

Condition a galvanic skin response to – e.g. Chicago, then:

GSR evoked by Chicago unattended message, though P does not notice/remember Chicago (Corteen and Wood, 1974)

Effect generalises to related city names – i.e. meaning has been activated, not just sound pattern detected

But GSRs to unattended weaker than to attended names: Semantic activation by unattended input attenuated though not blocked

Question 6

‘late selection’ theories

Answer 6

Theory: Both attended and unattended words processed up to and inc identification and meaning activation; relevant meanings then picked out on basis of permanent salience/current relevance

But: Doesn’t explain why:

selection on basis of sensory attributes so much more efficient than selection on basis of meaning

GSR to unattended probe words weaker than to attended

Question 7

what is the filter-attenuation theory?

Answer 7

Treisman (1969)

Early filter but:

Not all-or-none: Attenuates unattended sources – hence, with support of top-down activation, unattended words, if salient/contextually relevant, can still activate meanings

Early filtering optional strategy (not fixed structural bottleneck)

Question 8

why is an early selection option not a structural bottleneck?

Answer 8

Monitoring for target word (animal name)

Left ear: plate, cloth, day, fog, tent, risk, bear

Right ear: hat, king, pig, groan, loaf, cope, lint

Task: press L or R key when animals name heard

After practice, target detection as accurate when word target must be detected on either ear as only on one (Ostry, Moray and Marks, 1976)

Unless selective understanding/repetition of one message required

Question 9

what is inattentional blindness?

Answer 9

Exps by Daniel Simon require P to attend closely to one coherent ‘stream’ of visual events on screen, spatially overlapping with another stream

Highly salient events in unattended stream missed by large proportion of Ps (50% in exps)

Events of unattended stream, though happening in part of visual field, fixated by P don’t appear to be processed to level of meaning

=Early selection

Question 10

what is the attentional spotlight (of covert attention)? (Posner et al., 1978, 1980)

Answer 10

‘Endogenous’ cueing

Probable stimulus location indicated by arrow cue (80% valid)/ not – neutral cue – P responds as fast as pos (maintaining central fixation)

Simple RT to onset

Choice ‘spatial’ RT 9above/below centre)

Choice ‘symbolic’ RT (letter/digit)

All faster for expected location and slower for unexpected)

Question 11

endogenous v exogenous shifts

Answer 11

RT faster after sudden onset/change at stimulus location

Although does not predict stimulus location

Timing of ‘exogenous’ cueing diff from endogenous

Exogenous attraction of ‘spotlight’ fast (<200ms)

Endogenous movement of ‘spotlight’ takes several 100ms

Question 12

early selection in primary visual cortex and even LGN (O’COnnor, Fukui, Pinsk, Kastner, 2002)

Answer 12

Retina  lateral geniculate nucleus  V1

While fixation maintained on central point:

Series of digits appears at fixation and

High/low contrast checkerboards appear in left and right periphery

P either counts digits at fixation/detects random luminance changes in left (/right) checkerboard

fMRI BOLD signal in LGN/V1 voxels that react to checkerboard luminance change greater with attention directed to that side (red) than with attention to fixation

At least some selection (for regions in visual field) occurs v. early in processing

Question 13

visual selection (as for auditory attention)

Answer 13

Not all-or-none: Gradient of enhancement/suppression across visual field

Optional process: Size of attended area under voluntary control

E.g. on most trials, diff groups of subjects classified:

Central latter or

Whole word (proper word v not)

But responded also to occasional probe displays: Classify as & v T or Z

Question 14

does the efficiency of early selection depend on processing load?

Answer 14

Lavie (1995)

‘Flanker’ task: Press : for little x and right for little z on midline, ignore big letter above/below mid-line (distracting flanker)

Question 15

visual attention

Answer 15

Processing of info in visual field to level of recognition and meaning highly selective and limited:

‘Spot-light’ of visual attention can be moved voluntarily (and slowly) to locations/objects of potential interest away from fixation – attracted automatically (and fast) to local transients in visual field – size of spotlight can be varied/’zoomed’

Selective filtering occurs early in processing of visual info

Processing of objects outside spotlight relatively shallow (little ev for object recognition/activation of meaning)

Question 16

what are the functions of selective attention?

Answer 16

Defensive filtering: Protecting higher-level limited-cap systems (e.g. syntactic parsing) from overload (Broadbent)

Pos selection-for-action: Prioritising one of several possible objects of sources for action of further processing (Allport)

Feature integration/’binding’ (Treisman and Gelade, 1980) – combining properties of object analysed in diff cortical ‘maps’ – ev:

Visual search for conjunction v feature targets

Illusory conjunctions for unattended objects

Question 17

what is the feature-integration theory?

Answer 17

Regions of visual cortex are specialised for the local analysis of different attributes: colour, form, movement, texture, etc: each provides a “map” of that attribute’s values over the visual field

A discontinuity within one of these maps is detected automatically: i.e. a singleton onset/change/odd-one-out “summons” attention. Hence “pop-out”: no effect of number of objects with disjunctive feature target.

To bind features of the same object from different maps into an object description, we need focal attention to a location

Implication 1: to find a target identified only by conjunction of features (e.g. red + square in background of other red or square objects) will require focal attention to each object (or small group) in turn. Hence serial search required.

Implication 2: Features are “free-floating” in relation to each other until attended. Hence “illusory conjunctions.”

Treisman and Gelade (1980)