Cognitive Psychology - The processes involved in attention.
Cognitive Psychology (PSY323M1)
Lecture 4. The processes involved in attention.
Reading: Chapter 5 Eysenck & Keane; Styles, E.A. (1997). The psychology of attention. Psychology Press, Hove
Cognitive Lab: Eriksen task in Superlab folder
Your objectives are to-:
i). Understand the notion of attentional selection and attempts to pinpoint the location (i.e. 'early', 'late' or 'flexible') of the attentional filter in the information processing chain.
ii). Assess the utility of the different metaphors that aid our understanding of visual selective attention.
iii). Understand current thinking regarding the fate of 'unattended' stimuli, and in particular the results derived from the 'negative priming' paradigm.
iv). Be cognisant of the processes involved in visual search, with particular emphasis on Treisman's (1980) Feature Integration Theory.
v). Be able to describe some of the major disorders of attention.
When you are having a conversation in the coffee bar, a place that is notoriously noisy, you are most often so engrossed that you remain oblivious to the noise all around you. Likewise, when you are looking for someone in a crowd, especially if they are well known to you and the crowd isn't too large, it is most often a relatively effortless task to pick them out. How do we pick out and concentrate on that one sound source from among all others as we track a conversation? And how do we manage to identify that one object in space from among all the other objects? That is, what mechanisms or processes govern our ability to focus our attention on one source of information and exclude all the other sources?
Of course, we do not always have the luxury to pay full attention to a conversation because as often as not we are required do other things at the same time - like drive a motor car, negotiate our way along a crowded pavement or find that friend amidst the crowd. So how do we manage to divide our attention so that we can carry out multiple tasks at the same time? The fact that it is a normal human activity, indeed an expectation of all but the most intellectually feeble, is captured rather well in Lyndon Johnson's denouncement of Gerald Ford as someone 'who couldn't fart and chew gum at the same time'!
Focussed Attention
An early worker in this area, Cherry (1953), presented different messages simultaneously to each ear, and asked subjects to repeat the message they heard in one ear - a task known as 'shadowing'. He found that they remained unaware of the fact that the ignored message was actually in a foreign language or in reversed speech. Broadbent (1958), carried out a number of similar studies using a dichotic listening task in which a different series of digits were presented simultaneously to each ear. So for example, 2 was presented to the left ear and 5 to the right, then 9 to the left and 3 to the right, 1 to the left and 4 to the right and so on. The results showed that rather than recall the digits in the order in which they were presented ( 2,5,9,3,1,4 in this example), subjects recalled by ear of presentation (2,9,1 and 5,3,4 in this example). These results led Broadbent (op cit.), to propose that all stimuli are registered in parallel, and at a very early stage in the information processing chain stimuli are selected, or allowed to pass through a filter, where further processing of the selected stimuli takes place. Stimuli that are not selected may be held in a buffer for possible later processing. The physical characteristics of the stimuli provide the basis for the selection process. So on both Cherry's (op cit.) and Broadbent's (op cit.), task the basis of selection was the ear of presentation. The advantage of an attentional system organised in this way is that the capacity limited mechanisms deployed to further process information (e.g. mechanisms involved in memory or language), are not overloaded. The ignored information in Cherry's (op cit.), study did not receive any further processing so they remained unaware of the nature or content of that information. This is an early selection account of attention and it may be summarised as follows -:
A. All stimuli gain access to sensory mechanisms and are briefly held in store.
B. Where there is more than one message, a selection is made on the basis of the physical characteristics of the message and it is passed on to other mechanisms for further processing. The ignored message is held in store until the system is ready to handle it, if that is required.
C. This filtering process is necessary so that the limited capacity part of the system doesn't become totally overwhelmed by the available information.
However, later work challenged this account of attention. For example, Allport, Antonis and Reynolds (1972), presented a passage of prose to one ear and a series of words to the other ear. The subjects task was to shadow the passage of prose. They found that recall of the words was very poor. In contrast if subjects were presented with pictures, rather than words, whilst shadowing the passage of prose, they were later able to identify up to 90% of them. This suggested that the filter may not be as rigid as originally proposed and that if the stimuli were sufficiently dissimilar (in this case visual versus auditory) the information in the non-shadowed channel may indeed be processed.
Also, Von Wright, Anderson and Stenman (1975), performed a two-stage experiment in which words were firstly presented to subjects. Some of the words were accompanied by an electric shock. In the second part of the experiment subjects were required to perform a shadowing task in which these words, together with words that had not been associated with shock, comprised the stimuli for the 'ignored' message. They found that the shock associated words were associated with an increased galvanic skin response, suggesting that the words had not been filtered out during performance of the shadowing task. Words with a similar meaning or similar sound produced the same effect suggesting that attentional selection may occur quite 'late' in the information processing chain. Results like these suggested to Deutsch and Deutsch (1963), that attentional selection occurs after the identity of stimuli have been established. Their 'late selection' account of attention can be summarised as follows -:
A. All stimuli are fully analysed.
B. The selection bottleneck occurs near the response end of the chain of processing.
C. The most important or relevant information is the determinant of the response.
Treisman and Geffen (1967), asked their subjects to perform a shadowing task, but to tap a pencil on the table if they heard a target word in either message. If late selection theory was correct, then since the identity of all stimuli were established before selection, the subjects ought to have been as good at detecting the targets on the non-shadowed as on the shadowed ear. However the results showed that only 8% of the targets on the ignored channel were detected as opposed to 87% on the shadowed channel. These results provided strong evidence in favour of Treisman's (1963), model of attention which proposes a much more flexible location for the 'bottleneck'. Her theory can be summarised as follows -:
A. The processing bottleneck is much more flexible.
B. Information is processed through a hierarchy in which the information is first analysed in terms of its physical characteristics, then syllabic pattern, words, grammatical structure etc.
C. If the system encounters capacity-limits then the information is not fully analysed all the way to the top of the hierarchy, but stops at the point where further processing is no-longer viable.
The Eriksen Paradigm
The early work in the field of attention was concerned mostly with auditory attention. Eriksen and his associates investigated visual spatial attention. The task he used for the purpose required subjects to respond, in the simplest case, to two stimuli. For example they had to press button one if the target was an A and button 2 if it was a B. The subjects were cued as to where on a computer screen the target would appear. Most often an asterisk was displayed in the middle of the screen for 500ms so that the subjects could fixate on that point before the target letter was presented. So the task was a very simple one to perform.
However, the target was not always presented in isolation. On some trials the target was accompanied by distractor letters presented adjacent to the target on either side. What he found was that the exact location of the distractors, and whether they were the same or different from the target, affected subjects reaction times. If the distractors were the same as the target (e.g. in the compatible condition, A A A), reaction time was not affected. If the distractors were different from the target (e.g. A B A, the incompatible condition), then reaction time was slowed, but only if ...
This is a preview of the whole essay
However, the target was not always presented in isolation. On some trials the target was accompanied by distractor letters presented adjacent to the target on either side. What he found was that the exact location of the distractors, and whether they were the same or different from the target, affected subjects reaction times. If the distractors were the same as the target (e.g. in the compatible condition, A A A), reaction time was not affected. If the distractors were different from the target (e.g. A B A, the incompatible condition), then reaction time was slowed, but only if the distractors were quite close to the target. In fact the slowing effect only occurs if the distractors are within one degree of visual angle of the target. Outside of that area incompatible distractors have no effect.
This led Eriksen (1974), to propose that attention is like a spotlight so that anything within the 'beam' is selected and processed. Those stimuli lying beyond the beam's boundaries are not selected. His theory therefore assumes an early selection account of attention in which a physical attribute of the stimulus - its location in space - determines selection. He further proposed a 'continuous flow' conception of the attentional system in which activation of the mechanisms responsible for issuing the response is increased as information as to the identity of the stimulus accrues. If incompatible stimuli lie within the beam of the spotlight then two possible responses are under preparation, and it is the resolution of the competition between these that accounts for the increased reaction time.
Negative Priming
Until relatively recently attention was thought about in terms of the activation of processing mechanisms on selected stimuli. Whether early or late the assumption was that if stimuli were not selected then they received little or no processing , so that their fate was of little consequence. However, Tipper and his associates have demonstrated the phenomenon of negative priming which suggests that the process of 'ignoring' one stimulus in favour of another is an equally active process as is selection. In the typical experiment a picture that must be so ignored on one trial becomes the target on the next trial. In comparison to control stimuli the reaction time to such targets is significantly slower. The conclusion drawn by Tipper is that the activation level associated with the representation of the stimulus has been lowered as a consequence of being 'ignored'. So it is not the case, at least in this paradigm, that ignored stimuli are simply stimuli that have not received further processing.
Visual Search
Let's go back to that face in the crowd again! How do we pick out the person we have been looking for from among all the other people? How do we go about finding that bunch of keys from among all the clutter around and about? In other words, what is it about the relationship between different stimuli that determines selectability, and that makes selection relatively effortless on some occasions and much more difficult on other occasions? In order to answer these questions much use has been made of visual search tasks.
In the typical visual search task a target must be detected among distractor items. For example, deciding whether a red letter is present in a display comprised of all letters of the alphabet. As you can imagine the task is easy if all the non-target letters are blue, so you might expect to perform this type of task quite quickly and efficiently and wouldn't be much affected by the number of distractors in the display. A single attribute - the colour red - determines selection in this case. However, if the target was a red O and the distractors were comprised of red X's and blue O's what would the effect be on your time to decide whether the target is present or absent? In this case a conjunction of two attributes determines selectability - the colour red conjoined with the O-shape. Treisman and Gelade (1980), found that subjects' decision times increased linearly with display size when performing such a task, but that they remained unaffected by display size when selection was determined by a single attribute. Treisman (1988), proposed that focussed attention was not required to detect the target when a single attribute determined selection, but was required when selection was determined by the conjunction of attributes. Treisman's (1988) Feature Integration Theory can be summarised as follows-:
. Before attention is deployed the individual features of objects are subject to rapid parallel processing.
2. The process whereby individual features are combined to form objects is carried out by attentional mechanisms in a serial process which is much slower than the initial parallel process.
3. When focused attention is directed to the location of the object it acts as a 'glue' to combine the individual features together.
4. The process by which features are combined together to give a representation of the object is affected by the knowledge and experience we have of that object.
5. If attention is not properly focused, or relevant stored knowledge is not available, then 'illusory conjunctions may be formed.
One way of preventing attention becoming focussed is to present the display for a very short duration. In such conditions subjects report conjunctions of attributes that were not actually present - 'illusory conjunctions'. For example, if the target was a red O and the distractors comprised red X's, blue O's subjects might report seeing blue X's. This suggests that the individual attributes have been properly registered, but that the later attentional processes were not sufficiently focussed to combine the attributes appropriately.
Duncan and Humphries (1992), suggest that a simpler account of visual search performance if one focusses on the similarity of the target to the distractors and the dissimilarity among the distractor items. The greater the similarity of the target to the distractors, the greater the slowing of performance. And the greater the dissimilarity between distractor items the slower is performance. The major tenets of their theory are -:
i). All stimuli are processed at an initial parallel stage in which major groupings and segmentations are analysed.
ii). Attention acts to select and transfer selected information into short-term memory.
iii). This selection process is fast if the target is dissimilar to non-targets, and slow if they are similar.
iv). Likewise, if non-targets are dissimilar to each other, they cannot be rejected as a group and will therefore slow the selection process.
Disorders of attention.
Robertson, Lamb and Knight (1988), have shown that there are hemispheric differences in attention, such that patients with right hemisphere lesions have difficulties with global aspects of attention, while those with left hemisphere lesions have greatest difficulty with attention to local detail. A condition known as 'unilateral visual neglect' is associated primarily, although not exclusively, with right parietal lesions.
A patient with unilateral visual neglect is unable to attend to stimuli located in one side of visual space. So, for example, the person might often bump into objects on one side but not the other, or eat from only one side of the plate. When asked to make a copy of an object their representation will often exclude features of one side of the object; or if asked to draw the face of a clock will include all of the digits on one side. For some reason their copy does not seem odd to them. These patients are not blind in one region of space, so their problem stems from attentional 'neglect' of one side of visual space.
Using an attentional cue to orient attention to one or other side of space Posner (1980), found that the cue facilitates target detection among control subjects when it is valid (i.e. the target actually appears in the cued location) but hinders it when the cue is invalid (i.e. the target appears in a non-cued location). Neglect patients had little difficulty in performing in the valid condition, but their performance was much worse than controls in the invalid condition. In explaining his results Posner (1988) proposed that visual attention can usefully be understood in terms of the ability to engage a stimulus, the ability to disengage from a stimulus and the ability to shift attention.
Since patients had no great difficulty in the valid condition, Posner (op cit.), argued that their ability to engage stimuli was not affected. Neglect patients can also successfully shift attention. But performance in the invalid condition showed that when the cue was to the non-neglected side and the target to the neglected side visual attention could not be disengaged to be re-directed to the neglected side of space. Other studies (e.g. Rafal and Posner, 1987), have demonstrated that patients with thalamic lesions display difficulties in engaging attention to the space contalateral to the lesion.
A problem closely associated with unilateral visual neglect is visual extinction in which patients with lesions to the parieto-occipital cortex exhibit no difficulty in identifying single objects, but can 'see' only one when two are presented simultaneously. Patients can name an object presented to their 'bad' side, but only if nothing has been presented simultaneously to the 'good' side. It seems as if the presence of a stimulus in the 'good' field extinguishes the response to the stimulus in the 'bad' side. Volpe et al (1988) suggest that the stimulus in the 'bad' side receives a high level of processing, but that the level is not sufficient to support full conscious awareness.
Divided Attention
Often in our everyday lives we perform more than one task at a time. For instance we walk and talk, or drive and listen to music, and most often perform these activities successfully. However, we are often very aware of the difficulty or impossibility of trying to do too many things at once, and the consequences of attempting to do so. In many ways the study of divided attention is aimed at mapping out the limits of the information processing system. For the most part work in this area has focussed on the performance of two tasks at the same time, or dual-task performance as it is known.
Dual-task performance
Practice on one or both of the tasks, the similarity of the two tasks and the level of difficulty of the tasks are the main features that are held to account for our ability to perform dual tasks.
Spelke et al (1976), have shown that with sufficient practice it is possible to successfully read and write to dictation at the same time. Allport et al (1972) report similar findings with piano playing and shadowing and Shaffer (1975), with typing and shadowing. So it is possible with long periods of practice to achieve successful performance on a pair of highly demanding tasks.
Where two tasks are very dissimilar, like driving and holding a conversation, task performance is often as good as the performance of either task on its own. However, the more similar the tasks the greater the difficulty. But what does 'similar' mean? Wickens (1984), distinguishes three types of task similarity - tasks that make demands on the same modality (e.g. both require visual processing); tasks that make demands on the same stages of processing (e.g. input, translation or output stages); tasks that require the same memory codes (e.g. visual or verbal).
Task difficulty can be defined in many ways as you might expect. For example a reasonable person would agree that a task requiring the occasional memorisation of a list of three digits is easier than one were long lists of digits must be memorised. An important point, however, is that the difficulty experienced in the performance of a dual task is not simply the sum of the level of difficulty associated with each task independently.
Theories of dual-task performance
Bottleneck theories:
Evidence for a bottleneck in performance comes from the phenomenon of the psychological refractory period. That is, once a response has been made to a stimulus the cognitive system is momentarily less efficient at responding to a second stimulus (Welford, 1952; Pashler, 1993).
Central capacity theories:
These theories postulate that we have a single pool of undifferentiated resources and that performance of a task makes demands on these resources. In the dual task situation the degradation of performance on the secondary task occurs because there are insufficient resources to do so due to the demands of the primary task (Johnson & Heinz,1978; Kahneman 1973; Norman and Bobrow, 1975).
Modular theories:
A number of authors have proposed that modules of highly specific processing resources are deployed in task performance (e.g. visual, verbal, motor), and that difficulty in performance is due to competing demands for the same specific resources (Wickens, 1984).
Automatic & controlled processing
Shiffrin & Schneider (1977) and Schneider & Shiffrin (1977) propose that there are two qualitatively different processing modes that mediate our performance of different tasks. These are automatic processes that are held to be fast, capacity-free and executed in parallel. Additionally, they operate below the level of conscious awareness, do not require attention and are unavoidable. They are also considered to be difficult to modify in changing circumstances. Controlled processes, in contrast ,are slower, require attention and are capacity-limited, but are responsive to changing circumstances.
Conditions:
Consistent mapping -: memory-set targets never used as distractors over trials.
Varied mapping -: memory-set targets sometimes used as distractors in subsequent trials.
Memory-set: the to-be-remembered targets (1-4).
Display-set: the stimuli displayed during the test phase.
Results: The reaction time to targets is fast and unaffected by memory-set size or display-set size in the consistent-mapping condition.
Reaction time to targets is slower and is affected by both memory-set and display-set size. The larger the memory-set, the slower the reaction time; and the larger the display-set, the slower the reaction time.
Problems:
That there are upper and lower levels of control (Broadbent, 1971), or controlled and automatic (Schneider and Shiffrin 1977), is not a contentious issue. Early work examining skilled performance suggested that with practice a shift took place from a 'cognitive phase' to 'automatisation' (Schmidt, 1968), or from 'open' to 'closed loop' control (Keele, 1973). More recently much work has been carried out within the automatic/controlled processing paradigm which posits two distinct and qualitatively different processing modes. There is general agreement that controlled processing is slow, serial and capacity limited, so the greater focus of investigation has been on the charactersitics of automatism, particularly in regard to performance on simple visual search tasks. These range from Jonides (1981) suggestion that on engagement with a stimulus an automatic process runs to completion, to Posner's (1978) notion of 'invariance', and Schneider and Fisk's (1983) assertion that automatic processes occur in parallel and without effort.
The notion of resource comsumption lies at the heart of the automatic/controlled distinction. Resource consumption is an appealing but limiting conceptualisation of control. It is based on Kahneman's (1973) idea that task performance draws on an undifferentiated pool of resources. Thus evidence from dual task studies that suggest no mutual interference are interpreted as one or both tasks not requiring resources (i.e. automatic), while when they do interfere the result is interpreted as demand exceeding the supply of resources (i.e. controlled). As Allport (1980) has pointed out, this level of explanation is empirically unfalsifiable. The automatic/controlled distinction therefore uses the single pool assumption as the basis for distinguishing between levels. However, since only a single criterion is used the paradigm can only dichotimise between the two modes. Additionally, since only a single criterion is used there are no constructs available on which the two levels can interact. These are important to account for the delegation of control with practice to lower levels.
Delegation
This refers to the processes by which 'attentional withdrawal' (La Berge, 1975) takes place with practice on a task. These, and the observation that skill is organised into larger units over time, are not well understood. The A/C paradigm contributes little to our knowledge of them, since it focusses on outcome rather than process, laying down the conditions for the production of automatism (consistent-mapping of responses to stimuli) - not why, or by what mechanisms, it is produced. Several studies (Shiffrin et el 1981) have reported a deficit in performance following the combination of automatic and controlled tasks, a result from which it can only be concluded that the supposed automatic element of the task was in fact not automatic at all. A later dual-task study by the same authors found improved consistent-mapping performance, but since it could not be claimed that automatic processes had become more automatic it was suggested that subjects had learned to 'integrate' the two tasks. Integration is precisely what the A/C paradigm cannot account for. On a related theme Duncan (1986) questions the interpretation of complex behaviours such as playing tennis, driving and emotional disputes in terms of automatic and controlled processing. He suggests that in a simple task a response may be automatically produced to a single stimulus in a display, but that it is doubtful if in the real and complex world the unit of stimulus description is so easily identifiable. Discovering the principles that define the level or unit size of stimulus events that allow 'automatic' processing is the key. The study of prolonged practice ought to aim at the identification of the parameters which allow behaviour to be unitized and delegated to lower levels.
Intervention
Control can be conceptualised as an executive process intervening in other processes, or paradoxically as control being demanded (e.g. in the Stroop task, or an 'automatic attention response'). On the A/C model there is no means from the automatic level up to the controlled level. Control is conceptualised as a function of the highest level in the system and as unidirectional - down from the controlled to the automatic. This lies at the heart of the difficulty it finds in explaining how an automatic, attention-free process can paradoxically interfere with upper level attention demanding processes.
Automatic activation
The Control of Action
For Norman and Shallice (1980) action is controlled at its lowest level by ensembles of memory schemas, or knowledge units. Well-learned actions are mediated via a group of schemas in which the source schema acts, like a godfather demon, as the highest level of control and through which others are activated. Schemas are held to have different states -:
Dormant - these are located in permanent memory and are not involved in ongoing activity.
Activated - these are in a state of readiness for performing a function, but remain below threshold. Increased activation has several sources -
a) the source schema
b) deliberate attention
c) motivation
d) other activated schemas
e) the degree of match/mismatch between
trigger conditions and the trigger data-base.
Selected- these have an activation level that exceeds threshold and control action, both processing within the internal environment and the manipulation of effectors. Repeated use lowers the activation threshold. Once selected a schema continues to operate regardless of any change in activation value until the goal has been achieved, or it is blocked from utilising a resource by a more highly activated schema.
Horizontal Threads - schemas form a horizontal thread of processing structures. Numerous threads are held to be active at any one time in the control of actions, forming other schemas and operating on a memory data-base. Simultaneously active threads might conflict for the same resources or produce conflicting actions. Three devices are proposed to resolve such conflicts - vertical threads, contention scheduling and triggering conditions.
. Vertical Threads - these are comprised of attention and motivation. The former acts via the supervisory attentional mechanism which monitors ongoing activity and increases or decreases activation values. It modulates activity, so it does not directly select a schema but creates a bias. The mechanism has a limited capacity and on occasion this may be exceeded. Experience provides refinement of schema activation values, allowing the establishment of horizontal threads without vertical thread influence. Motivation acts over a longer time period, affecting schema activation values so as to bias them in accordance with long-term aspirations.
2. Contention-scheduling - this involves the lateral inhibition/ excitation of schema activation. By this means schemas compete for selection and co-operate in the use of common processing structures. The mechanism ensures that only one schema is selected at any one time.
3. Trigger Conditions - this element of the theory solves the problem of the fine-timing of actions to coincide with events in the environment. Each schema has a set of trigger conditions and the match/mismatch between trigger and the trigger data-base forms the third selection mechanism.
All three of these conflict resolution devices act on the simple variable of schema activation value. This prevents incompatible schemas from being selected and allows for the operation of many.
Evidence:
Multiple targets - in this situation performance is crucially dependent on the establishment of the appropriate trigger conditions. These may be difficult to establish, for example, in a varied mapping paradigm. Also, once established, trigger conditions may permit the inappropriate selection if distractor stimuli are similar to targets.
Dual tasks - two conditions must be satisfied for successful performane in this situation. Firstly, horizontal threads must be sufficiently well-established to permit control without input from the supervisory attentional mechanism. Secondly, selected schemas must not give rise to structural interference. This is achievable if the overlapping use of common processing mechanisms is minimised.
Other experimental evidence-
Shiffrin's hair-trigger response basis following a preview.
Facilitatory and inhibitory effects of semantic priming on response latency.
Automatic and controlled processing.
Phenomenological evidence
Automatic performance with practice: Much of our behaviour becomes automatic with practice, lots of sub-actions being controlled by schemas with highly tuned triggering conditions. Large segments of behaviour are thus performed without supervisory attention so that the control of action resides at a low level over relatively long time periods.
Automatic performance and interference: Well-learned tasks do not interfere because there is less need for supervisory support, triggers are more specific and schemas are more specific in their demands on processing structures.
Neuropsychological evidence
Frontal lobe lesions: The frontal lobes have traditionally been viewed, since Luria, as being involved in planning, monitoring and verification. Lhermitte et al (1972) report a case study in which a frontal patient is asked to solve a relatively simple task, but one that requires a little planning - ' The son is 15 years old; his father is 25 years older; his mother is 5 years younger than the father. What is their combined ages?' The patient scored normal on digit span, picture completion and the similarities sub-test of the WAIS, but in response to this question added 15, 20 (a transcription error) and 5, producing 40 as the 'total age of the father, son and the holy ghost'. Shallice (1988) suggests that this type of behaviour is typical of frontal patients and results from poor monitoring, selection and control. He reports the case of an accountant tested six years after injury who had an I.Q. of 130 and performed satisfactorily on a wide range of tests. The patient's life was disastrous having gone bankrupt, lost several jobs, divorced and remarried. His employers complained of his lack of punctuality and his general disorganisation, while his wife reported that he took hours deciding where to go for dinner, discussing seating plans, menues, atmosphere, management, even driving driving along to see a possible venue to check how busy it was, etc., and still couldn't make a decision.
It is pointed out by Shallice (1988) that two paradoxical predictions are made by the control of action theory and that each receives support from neuropsychological patients. The first is that contention-scheduling mediates routine action and is dependent on the reinforcement history of the organism. Without the aid of the supervisory mechanism this system may produce behavioural rigidity, or perseveration. It is often the case that frontal patients perseverate in performing the Wisconsin card-sorting task, being unable to change the rule. The second is that if a stimulus situation does not produce strong trigger-schema activation, then in the absence of the supervisory system, contention-scheduling may be frequently captured by irrelevant stimuli. An example of this sort of behaviour is observable in 'utilisation behaviour' in which patients continually use objects that have been placed in their immediate vicinity.
Is the control of action theory testable?
• The benefits of having a broad set of theoretical constructs that are meaningfully related for the explication of a large body of evidence must be contrasted with the benefits of specific theories for specific effects.
• Interpretative frameworks.
• Top-down versus bottom-up explanations.
• Testing parts of the model.
• Interfacing different bodies of literature (e.g. experimental with neuropsychological)
Automatic process & automatic retrieval
Logan's (1988) theory can be summarised as follows-:
. The memory trace of a stimulus includes information about it, including the response made.
2. With practice the amount of information is built-up, including the most oft-used or appropriate response.
3. Eventually, the stimulus is accorded a 'single-step, direct access' to the response.
4. Automaticity is thus achieved as part of the retrieval process so that past solutions can be immediately enacted.
Action slips
Reason and Mycielska's (1984) diary studies.
Ecological vailidity:
• concerned with long chains of behaviour in the real world rather than with discrete responses in laboratory tasks.
• seems sensible and useful to examine 'real' behaviour for clues as to how the cognitive system is organised rather than progress within the constraints of a particular paradigm.
On Reason and Mycielska's (1984) model the 'Intention System' can be captured by internal preoccupation or by external events. Thus focal attention may wander onto parallel mental activity so that the critical choice point in a sequence of action is missed and a well-established sequence is run through to completion, or an unexpected stimulus draws the intended sequence onto an unintended route. For example -:
. 'I had decided to cut down my sugar consumption and wanted to have my cornflakes without it. However, I sprinkled sugar on my cereal just as I had always done.'
2. 'We now have two fridges in our kitchen and yesterday we moved our food from one to the other. This morning I repeatedly opened the fridge that we used to have our food in.'
3. 'On starting a letter to a friend I headed a the paper with my previous home address instead of my new address.'
4. 'I intended to stop on the way to work to buy some shoes, but woke up to find that I had driven right past.'
5. 'I brought the milk in to make myself a cup of tea. I had put the cup out previously. But instead of putting the milk into the cup, I put the bottle in the fridge.'
6. 'I meant to get my car out, but as I passed through the back porch on my way to the garage I stopped to put on my Wellington boots and gardening jacket as if to work in the garden.'
7. 'I have two mirrors on my dressing table. One I use for making up and brushing my hair, the other for inserting and removing my contact lenses. On this occasion, I intended to brush my hair, but sat down in front of the wrong mirror, and removed my contact lenses instead'.
8. 'I went up to my bedroom to change into something more comfortable for the evening, and the next thing I knew I was getting into my pyjama trousers'.
9. 'I meant to take off only my shoes but took my socks off as well'.
0. 'I was making shortbread and decided to double the amounts shown in the recipe. I doubled the first ingredient - butter - and then failed to double anything else'.
1. 'I decided to make pancakes for tea. Then I remembered we didn't have any lemons, so I decided not to bother. Five minutes later, I started getting together the ingredients for pancakes having completely forgotten my change of mind'.
2. 'I was putting cutlery away in the drawer when my wife asked me to leave it out as she wanted to use it. I heard her, agreed, and yet continued to put the cutlery away'.
-3 All involve an intention to change from an old well-established routine. The old sequence intrudes.
4-5 Inattention contributes to exclusion rather than intrusion
6-7 Examples of branching errors in which attention is captured at the branch.
8-9 These are overshoots where the sequence is not stopped at the appropriate point.
0-11 A reversion to an earlier plan takes place.
2 Continuation of an earlier plan takes place.
These errors are caused, not by lack of skill, but by misapplied competence. They are errors of the highly skilled, not of the learner, and tend to occur where sequences are similar and where one or other is frequently executed.
Some problems
* The claim that some errors are caused by internal and others by external factors is problematic.
* Self-selection bias
* Only those errors that are noticed are reported
* Potentially important factors may be forgotten or overlooked in the period between occurrence of the error and its reporting.
* Participation may change behaviour.
* Behavioural v conceptual error classifications.
* Mapping errors to cognitive mechanisms is problematic. The more errors, the more mechanisms or interactions between them are required to account for the behaviour. Any consequent theory is likely to be large and unweildy, and difficult or impossible to provide empirical verification/falsification. If there are too few error classifications any subsequent theory is assuredly incomplete.
Nevertheless, behavioural capture is a powerful concept and is well-supported by data - 40% of errors being classified as capture by independent raters. The original notion behind the construction of the Cognitive Failures Questionnaire (CFQ) by Broadbent et al (1982) was that errors may occur because of aberrations in the working of perceptual, memory or action systems. Subsequent studies failed to find any support for this notion, all of the CFQ items loading on a single factor. Work on Reason and Mycielska's (1984) Error-proness Questionaire (EPQ) recovered only two factors - memory and attention. The processes underlying capture may, therefore, be fundamental cognitive processes that are in operation over many other behaviours too.
3