A distinction is often made between attention and attentional set. ATTENTION is a decision process; the systematic admission of information into consciousness. ATTENTION SET is the capacity to selectively prepare our nervous system to process one set of stimuli, think about a topic, or make a response.

Attention is viewed as the process of selecting some of the many available inputs.

Cherry - dichotic listening/ shadowing task. What is coming in on the unattended ear? No meaning. Could tell it was speech, could tell if the voice was male or female.

Early Selection Theories:

Filter Theory - only one source of information is allowed to reach the stage of meaningful processing. Other information does not compete for the limited resources of higher level processing. The is an early selection model of attention. Broadbent postulated that attention is directed and maintained by sensory or physical attributes of the input message. Physical cues guide pattern recognition from the sensory register. So blocking occurs at the sensory level. While information is held in the SR it undergoes a Preattentive Analysis that determines its physical characteristics (pitch, intensity, gender,). The preattentive analysis determines what will undergo further processing or analysis. The limited capacity channel can become overloaded requiring switching (multiplexing). Evacuating information from the SR must be done serially, one channel at a time.

 Broadbent (1958) presented 3 pairs of digits; three to ear 1 and 3 to ear 2. Ss recalled digits ear-by-ear instead of switching from ear 1 to ear 2. (Ear 1 - 1,2,3,; Ear 2 -7,8,9) recalled 1,2,3,7,8,9 instead of 1,7,2,8,3,9

 Moray (1959) Ss shadowed messages on ear 1 while on ear 2 the same word was repeated 35 times. People still did not know what the word was.
Moray also showed that if the word on the unattended channel was the subject's name it was often recognized.

 Allport, Antonis, & Reynolds (1972) combined messages using verbal and pictorial information and found that the concurrent presentation did not interfere with attention .

 Treisman (1960) shadowing task; shadow what you hear in your right ear and ignore what is coming in on the left ear. She then presented a story to the right ear that continued in the left ear. Ss followed the story rather than the ear.
Conclusions: limitations on the processing of 2 concurrent stimuli are less rigid than implied by Broadbent and some meaning is coming in on the unattended channel

 Treisman Attenuation Model of attention.

Attenuation Model (Treisman, 1964) - attention acts like an attenuator turning down the volume on unattended channels as opposed to totally blocking them out. Allows gradations in the amount of information passing through. Attention becomes a matter of degree.Incoming stimuli undergo a series of tests which first analyze the inputs on the basis of physical characteristics (location, pitch, etc.) and then in terms of context. Irrelevant messages are heard with a dull ear not a deaf ear. Knowledge is responsible for tuning the filter.

 Grey & Wedderbrum  gave Ss a shadowing task (ear 1 - mice 3 cheese) (ear2 - 2 eat 4)

 MacKAy (1973) shadowed ambigious passages with information on the unattended channel that clarified the ambiguity (ear 1 - bank; ear2 - river or money). Memory test - Ss remembered sentences for which the unshadowed word gave meaning

 Moray (1970) conditioned GSR to certain words through shock. Presented the word or a synonym on the unattended channel.

 Incoming information might undergo three tests:

1)  physical property analysis
2)  linguistic analysis (words/nonwords)
3)  semantic analysis (meaning)
 the analysis is carried out until stimuli are disentangled (e.g. only need a physical analysis to tell a woman's  voice from that of a man). Feedback from LTM can influence channel settings. Feedback from what has been processed switches you over to other channel.

 Cherry and Kruger (1983) selective attention in LD kids: Task: kids (7-9) required to point to the picture of a word that was represented in one channel. In the other ear, the kids heard either white noise, backward speech or forward speech.
  Findings: when anything came in on the unattended channel, LD kids performed poorly. Their performance was especially poor under the forward speech condition. On the other hand, normals were unaffected by the information on the unattended channel.

Problem: the attentuation model requires recognition of information prior to processing.
 Why propose that attention screens prior to meaning.

Late Selection Theories:

The Late Selection theorists propose that everything is meaningfully processed prior to pattern recognition. Selection is attention. Late selection allows for unconscious influences on attention.

 Deutsch and Deutsch (1963) Norman (1968) selection is assumed to operate on response output. All input activates a semantic representation; all information is recognized. However, we are limited in our capacity to respond to all input so we must select some part of the recognized input to respond to. The selection made is based on Pertinence.
  When someone says your name while you are attending elsewhere, the sensory input is still processed to the level of meaning. If this memory representation is coded as being important (your name is pertinent) then mental effort is allocated for further processing, so that you become consciously aware that your name has been spoken.  According to this model, stimuli are processed to the level of meaning automatically and there is a close relationship between memory and attention.
 

  Marcel (1980) asked Ss if a string of letters (e.g. brke) is a word (lexical decision task)
   Using word triplets :       hand  palm  wrist  (semantically consistent)
       tree  palm  wrist  (semantically inconsistent)
  The second word, Palm, was sometimes pattern masked. The experimental question was, what effect did the word Palm have on the decision time (is this letter string a word) Wrist.
  Findings: Palm facilitated RT for consistent triplets even when masked! The time to recognize WRIST as a word was faster when preceded by HAND  PALM whether masked or not. (greater when only HAND WRIST was shown)

  Further evidence for preconscious processing was the finding that under masked conditions only when the preceding words were TREE PALM, RT was facilitated.
  Marcel interprets these findings as support for the idea that the function of consciousness is selectivity. While all meaning of all words occurs at the preconscious level, only one meaning enters conscious awareness. Since all meanings of "PALM" are activated preconsciously in parallel, the preconscious meaning is availble to influence responding to WRIST.

Reaction to the Deutsch and Deutsch Model: According to Treisman, if all information is recognized, as the late selection model argues, then why are so few targets detected in the unshadowed channel /
Treisman & Geffen (1967) tested early versus late selection.
     Task 1: Ss heard 2 prose passages presented dichotically and shadowed one  Task 2: certain words were designated as targets. Any time a target word occurred in either the shadowed or the unshadowed channel, the Ss had to tap the table.
Findings: in the shadowed ear the target was detected 87% of the time ;In the unshadowed ear, the targets were detected 8% of the time.

 According to the  late selection theorists, the shadowing task is very attention demanding and limits our processing capacity at the selection level. In spite of this, 8% of the targets were detected. D & D shadowing is not a good test.

ERP's to examine the processing of unattended stimuli, Hillyard, Hink, Schwert, & Picton (1973) looked at response to tones presented to both ears. Ss were told to attend to tones in one ear and count the number of tones at a particular pitch. They were told to ignore the tones in the other ear. They found that the unattended tones were 20-70% smaller than the attended tones. Further ressarch found that scalp location near the auditory cortex showed maximal N100 responding, as well as maximal change in this response with attention - suggesting that attenuation is occurring early in auditory processing (Woldoroff & Hillyard, 91)
McCarthy and Nobre (1993) semantic processing of unattended stimuli - ERP response to a word includes an N400 response occurring at frontal or central electrode sites - thought to have to do with semantic processing because when a word is preceded by a prime, the response becomes significantly smaller (doctor- nurse). Using a visual divided attention task ,Ss were presented words in the right and left visual fields and were instructed to attend to one of the fields (while focusing on a central fixation point). They were told to respond when a target appeared in the attended field. Unbeknownst to the S, some of the nontarget words presented to both visual fields were preceded by a related prime. They found N400 responses for words in the attended field that is smaller for those words that were preceded by a prime - they did not find N400 response for words preceded by a prime in the unattended field, suggesting limited processing for words in the unattended field.
Conclusion: unattended stimuli are not processed to the same extent as attended stimuli, perceptually or semantically

Capacity Models of Attention (Kahneman, 1973) Kahneman - combines motivational, attentional, and arousal processes
This model argues that we have finite  processing resources or cognitive capacity to devote to a task and that different tasks require different capacity use. The number of activities which can be done simultaneously is determined by the capacity each requires.
Attention is the process of allocating capacity to various inputs.

 Norman & Bobrow (1975) performance in any situation is under two types of constraints: data limitations and resource limitations.

 Data limited tasks - difficult or impossible even if we devote all of our capacity; performance is limited by the quality of the data
 Resource Limited Tasks - refers to the quality or the level of effort used to accomplish a task. Performance can be improved by increasing the amount of processing . Performance is limited by the demands that the task makes on the cognitive system.

 Secondary Task technique and cognitive capacity - the amount of effort required by one task will come at the expense of capacity that can be devoted to other tasks. S is instructed to perform a primary task and a less important secondary task simultaneously. Estimates of the processing capcity required by the primary task are obtained on the secondary task; the longer it takes to react to the 2ndary task indicates the capacity required by the primary task. As task requirements on the primary task increase, performance on the secondary task declines.

 Theories of attention argue for the existence of three stages of perceptual processing:
  1) construction of sensory representations of input
  2) construction of semantic representations of these sensory inputs
  3) admits the products of earlier stages into consciousness
  Selection can occur at any one or all of these stages. According to Kahneman & Johnston & Heinz, selection consumes processing capacity and the amount consumed increases as one moves through the various attentional stages.

 Capacity (effort) - the amount of capacity allocated to a task is determined by  task demands and motivation

  2 aspects of effort:

1.  allocation policy (controlled by 4 factors):

a)  involuntary processes
b)  momentary intentions
c)  evaluation of task demands
d)  effects of arousal produced by external stressors

 2. evaluation of demands on capacity (causes effort     to be supplied to accomplish the activities that     the allocation policy has selected)

Tyler, Hertel, McCallum, & Ellis (1979) memory for words improves as the amount of cognitive effort devoted to learning the word increases. Words were presented as easy or difficult anagrams (CROODT or DORTOC; OTEALNEP OR ANLTEPOE). A secondary task occurred during the primary task (detect a tone) Results: Ss had slower reaction times to signals when working on more difficult tasks. Explanation: different capacity requirements

 

    Kahneman asserts that we have little control over the amount of effort expended while performing a task. The amount of effort is determined by the task (task determines effort); not by the amount of attention we decide to allocate.
 

 Johnston & Heinz (1978) amount of processing on the unshadowed channel varies as a function of task demands
 Ss listened to binaurally presented messages and were told to pay attention to one. The target was distinguished from the non-target by either: a sensory cue (different voices)  (male voice targets / female voice nontargets);
 a semantic cue or by both.In addition, a secondary visual detection task was performed. Ss were instructed to devote as much mental effort as needed to perform the listening task.
 Rationale:The more resources required to perform the listening task, the less residual capacity for signal detection
 Findings: Ss had quicker RTs in the sensory cue condition than in the semantic cue cond.. Suggests that selective attention is less effortful when two stimuli can be discriminated on the basis of physical or sensory cues alone.

 J & H pointed out that these findings could be interpreted as selection occurring at stage 3, so they did another experiment: in this experiment everything was the same as before except they varied the instructions. Ss were told to extract as much information as possible from both messages leading to a late stage 3 processing  in all conditions.
 According to D & D both messages are always processed to S3 irrespective of instructions, so the previously obtained differences should occur). Acc. to J&H, the diff. in expended capacity should only occur when an earlier mode of selection is used in the sensory vs. the semantic conditions. Since the new instructions require Ss to use late mode of processing, the difference in expended processing capacity should be eliminated - it was

 Johnston & Heinz (78) selection is possible at several different stages in processing; the more stages of processing that take place prior to selection, the greater the demands on capacity
 

 Johnston & Wilson (1980) presented pairs of words dichotically (one word to each ear)  Ss had to identify target items consisting of members of a designated category (targets were homonyms)  e.g. category - clothing / socks - smelly
                                clothing / socks - punches
                                clothing / socks - tuesday
 each word was accompanied by a non-target word that biased the appropriate meaning or the inappropriate meaning or was unrelated to the meaning of the target word (tuesday).
 Ss were either told that the targets would always be in the left ear or that the targets would occur in either ear.

 When Ss did not know which ear the targets would arrive at, appropriate non-target words facilitated target detection and inappropriate non-target words inhibited target detection

 Conclusion: when attention needed to be divided between the two ears (Ss did not know which ear would hear the target) , there was clear evidence that the non-target words were semantically processed. When the S knew that the all targets would be presented to the left ear, the type of non-target word had no effect on detection. These findings suggest that non-targets were semantically processed in the divided attention condition but not in the focused attention condition. The amount of attention given to non-targets is as much as is needed to accomplish the task.

 Rabbitt (1964,1967)    Ss required to locate a target letter among a number of irrelevant letters. Search time was initially affected by the number of irrelevant letters; but with practice, the interfering effects of the irrelevant letters was greatly reduced if the letters were unchanged (unless the new irrelevant letters shared the same features as the old irrel. letters) (straight lined letters)

Suggests that irrelevant stimuli are not processed fully; only essential features of the irrel. letters that serve to distinguish between targets and distractors were processed (similar to the different stages required by attention)

The ability to attend to a task is related to physiological arousal (the capacity model continues)

 Yerkes-Dodson Law (1908) as arousal level increases so does performance, up to a point; afterwhich increases in arousal lead to decreases in performance. Inverted U shaped curve

 Easterbrook (1959) applied the Y-D law to attention; as arousal increases so does attention, up to a point.....heightened arousal affects selectivity;
 STM (working memory) capacity is decreased with arousal; a progressive reduction in the range of cue utilization as arousal increases; reduction in attention to irrelevant cues leads to improved performance;
 when all irrelevant cues have been eliminated and with further increases in arousal, a further reduction in cues (tunnel vision) can affect attention to relevant cues thus reducing performance (start to ignore relevant cues)

 Task difficulty - inverse relationship between optimal level of arousal and task difficulty;

 More arousal causes greater amplification of the already activated units resulting in inhibition of the less activated units . If arousal is high, a few things are attended to and everything else is ignored

 Lateral inhibition (Walley & Weiden, 73,74) attention results from activation of a node that, in turn, inhibits surrounding nodes by means of lateral inhibition; the more activated the node, the greater the L.I.

 Physiological variables - heart rate and skin conductance seem to be associated with changes in attention allocation

 Concentration is viewed by most as the allocation of a large amount of mental effort to a particular task;
 it should be viewed as the ability to attend selectively.

-

-Multiple Resource Models (Navon & Gopher, 1979) separate resources or capacities - not a single capacity - both the amount and the type of resources required to perform a ttask are important in determining attention and performance (interference from similar tasks is greater)

Dawson & Schell (1982) hemispheric specialization; classical conditioning of words (word + shock), cc words presented in the unattended channel during a shadowing task
Varied ear of the unattended message
Left ear -( unattended )cc words led to autonomic responding but when Ss did not shift attention to the unattended ear there was no autonomic response for Ss who received the cc word in the right ear
Interpretation: ipsilateral projections (same-side); contralateral - more numerous and stronger in dichotic listening (competition between stronger contralateral and weaker ipsilateral messages
Right ear (attended) leads to left hemisphere (speech) leaving the right hem. Free to process the unattended message (left ear) where it analyzes the unattended message and reacts (autonomic response)
Left ear (attended ) leads to right hemisphere - to shadow the message the information has to go to the left hemisphere where the unattended message is also going leading to interference and overuse of the resources resulting in no response to the cc word

----------------------------------------------------------------------------------------------------------

Automatic versus attentional processing

 Automatic processing reduces the mental effort a task requires

 Attentional resources limited. Costs are reduced by practice
 Automatic attentional processes are fast, effortless, parallel, unlimited capacity, not under direct control

 Attentional or controlled processes are slow, serial, effortful, limited capacity, under direct control

 Neisser - all input activates LTM codes to some extent (Preattentive processing - a late selection view); attention equals full activation of LTM codes. Neisser refers to preattentive processing as automatic processing
 

 Posner and Snyder (75a,b) three criteria to determine if a process is automatic:
1)  it should not lead to conscious awareness
2)   it occurs without intention
3)   it does not interfere with other concurrent mental activity

 Used a Priming task to distinguish between automatic and attentional processing:
  Ss were told that a prime appearing before the test (0-1/2s) will be 80% useful or 20% useful. The task was to determine whether letter pairs were the same or different:
  Prime   Test   Correct response
     A     AA      same
     B     AA      same
     A     AB      different

 Neely (1977) the prime does not need to match the test stimuli to produce its effects: used a lexical decision task (does this string of letters form a word - Broni or Robin)

 Ss were presented with either an appropriate or an inappropriate prime (bird or body); findings: facilitation (faster decision time) with the  appropriate prime

 In a second study: he taught that a certain prime meant something other than its usual meaning ( BIRD referred to body parts/ BUILDING referred to birds)

 The effect of using these different category primes differed depending on how close the lexical decision task followed the prime: under short latency conditions - BUILDING-ROBIN vs. BIRD_ROBIN / Ss faster to respond to BIRD-ROBIN; under long latency conditions / Ss faster to respond to BUILDING - ROBIN

 Why?  takes approx.300-500msec for a conscious expectation to form
 

 We learn to process automatically:
 Shiffrin & Schneider and Schneider and Shiffrin (1977)

  task: to detect when a target occurs in a series of rapidly flashing visual arrays /consistent task: the memory set and the visual arrays are non-overlapping; target items are never seen as non-target items (targets are consistently targets)
   1) learn memory set (a,6,2,b,r,t)
   2) view arrays; 1) (e,r,s,3) 2)(g,s,2,n) 3) (m,c,8,j) etc.
varied task: the memory set and array items are mixed; a memory set on one trial might occur as a non-target item on another trial;
   trial 1 - learn memory set (a,g,l,h); array (b,r,c,a); trial 2 - learn memory set (b, k.n.z); array (a, m, r, k)
 Findings: for the consistent task; after practice S were able to search for several memory set items as rapidly as for one; 2100s - detection accuracy went from 55% to 80% at 120msec
   For the varied task (controlled search); Ss had to seek items in the displays demanding time and attention; under 400msec Ss performed at 75% accuracy (no learning occurred in the varied task)

 Automatic processes are inflexible and hard to modify (mental set in problem solving - 9 dots; habits; motor skills)
  Shiffrin and Schneider trained Ss on the consistent task; in this task the memory set items were always from the first half of the alphabet and array items were from the second half. Following many trials, reponding became automatic. Then, the criteria were shifted. Now, targets came from the second half of the alphabet and array items from the first. Detection accuracy dropped to <30% and it took about 3000 trials before performance reached the initial level

 

Selective Attention

 Selective attention  - task - to pay attention to one input in the presence of others (e.g. driving, searching for a product on the grocery shelves). Selective attention is sometimes easy and sometimes difficult:
  1) easy - when the target differs from other inputs at the level of separate dimensions
  2) difficult - when the target does not differ from the others on a single dimension
 

 Treisman and Gelade (1980) the isolated feature detection occurs automatically - pre-attentive processing; the combined features require controlled attention.

 Simultaneous detection - there are many innate perceptual limitations on the kinds of targets that we can detect simultaneously: Duncan (1980) when similar targets occur there is a greater likelihood of failing to detect one or the other than of failing to detect a single target; the same analyzers are activated signaling the detection of a single target and it takes time to reactivate for another location
 

 Visual dominance even though simultaneous inputs occurring in different modalities are detected more quickly than are similar inputs, detection is not perfect. 2 inputs are not subjectively perceived as having occurred simultaneously. If a light and a tone are presented simultaneously , the light is likely to be detected first. Tactile targets dominate over auditory targets. We give priority to the visual at the decision level. This arises from competition between inputs at the decision level rather than at earlier selective levels. This is referred to as the Law of prior entry. Using the ticks of the metronome to time the transit of stars they noticed that the star crossing was perceived before the tick.

 Divided attention - how much can we be consciously aware of in a limited amount of time?

  Procedure: determine how many of rapidly presented sequential elements you can count; 2 per second for visual items because of afterimages and 10 per second for auditory

  Psychological Refractory period - the period during which 2 responses interfere with each other Pashler (1994) when two cognitive tasks overlap, we are slower starting the second

Procedure: determine how many of rapidly presented sequential elements you can count; 2 per second for visual items because of afterimages and 10 per second for auditory

Psychological Refractory period - the period during which 2 responses interfere with each other

Example: playing basketball; players attend to several visual and auditory stimuli while dribbling the ball; how can he do it? Some of the behavior is automatic and much of the remainder is controlled by monitoring . Monitoring detects change (an economical use of capacity). Monitoring is demonstrated by the orienting reaction. Neural firing increases with novel stimulation and when repeated input is stopped or changed . In NYC people living over subways reported that their sleep was disrupted at 2AM one night. They couldn't figure it out. It turns out that the subway shut down for a short period causing silence; their automatic monitoring noticed the change.

What are the limits of divided attention? We talked before about driving and listening to the radio at one time and you all decided that you could under many circumstances do these things. What are the resource requirements of these activities? Driving? Talking? One is verbal and one is visual and motor; one requires motor outputs and the other requires verbal outputs. However, what happens when the traffic gets heavy? Therefore any resource required by both must be a general resource ( a resource applicable to a wide array of tasks). Concurrent task performance is easier if the tasks are distinct but some interference can also occur implying that there is some general resource required in the execution of the tasks. There are also task-specific resources and if two tasks share these resources, studying and watching TV (both verbal) then interference should be great. But the question remains - can attention be successfully divided on tasks that require many of the same resources? Can we divide attention between any two tasks?

Spelke, Hirst, and Neisser, 1976; Hirst, Spelke, Reaves, Caharack, & Neisser, 1980) gave Ss stories to read and while they read they wrote down dictated words. They were tested for comprehension. After 6 weeks of practice, Ss learned to combine the tasks. In the second study, the researchers found that the Ss also comprehended the dictated words. The words formed sentences and some of the recognition test sentences were "gist" sentences and the Ss falsely recognized them (heard - the dancers performed. The performance was excellent ; recognized - the dancers were excellent.)

Taken separately, these tasks - reading and taking dictation - are highly practiced routines, however, the practice needed was in coordinating these activities. It seems that it is practice with the combined tasks that matters rather than practice with the individual tasks.

Why is combined practice so important? Perhaps the limits to divided attention are not resource limits but are problems in bookkeeping - the chore of keeping straight which task is which. Remember, trying to rub your stomach and tap your head - the problem is one of channel segregation - keeping track of which hand is doing what - one suffers from cross talk between channels.
Shaffer (1975ab) used well-practiced typists; they were given printed material that they had to type while they shadowed auditory information through headphones - easy - however, when they had to read printed material and had to type the material coming through the headphones they couldn't do it. What are the task demands? The visual input belongs with the typing and the auditory input belongs with the shadowing. Channel segregation depends on being able to discriminate the channels.

 

Back to homepage