3.4.1 Cognitive control in the AB
As previously mentioned, this is not the first example of an RSVP experiment to find differences between participants regarding the AB effect. Martens et al. (2006) recorded electroencephalographic (EEG) activity from 11 blinkers and 11 nonblinkers who performed a dual target RSVP, and found that when T1 was successfully
identified nonblinkers showed more activity in the ventrolateral prefrontal cortex. This is an area linked to early target selection processes and they propose that if there is not enough activity in this area participants will have problems selecting relevant information from irrelevant information, resulting in interference in T2 processing. They postulate that the AB arises because participants fail to consolidate the relevant information fast enough; more activity is consistent with the early processing of T1, allowing T2 to be processed and therefore resulting in a smaller blink. This implies that individuals may differ in their ability to attend to task-relevant information and inhibit task-irrelevant information, potentially due to cortical activity (Martens & Valchev, 2009).
It is therefore equally plausible to suggest that the differences found may be due to differing levels of cognitive control; with some people having more difficulty controlling the allocation of attention than others. This notion fits with findings showing that the AB increases in magnitude and duration for people with attention- deficit hyperactivity disorder (Hollingsworth, McAuliffe, & Knowlton, 2001), people with moderate to severe dysphoria (Rokke, Arnell, Koch, & Andrews, 2002), children with specific language impairment (Lum, Conti-Ramsden, & Lindell, 2007), and older adults (Lahar, Isaak, & McArthur, 2001).
Taatgen, Juvina, Herd, Jilk, and Martens (2007) have associated these findings with differences in cognitive control. It is generally accepted that the more control an individual has in a specific task, the better they will perform. However when that task is a dual target RSVP this is not always the case and too much control causes a
detriment in performance, which is revealed through an increase in the AB. Taatgen et al. have found that the magnitude of the AB is correlated with performance on other tasks designed to measure cognitive control, for example, the abstract decision making task (Joslyn & Hunt, 1998, as cited by Taatgen et al.) in which participants must sort objects into bins once they have obtained enough information regarding the properties of the objects. They find that participants who show a larger blink also show greater levels of cognitive control in other tasks. An increase in cognitive control causes problems with tasks like the RSVP because it reduces cognitive flexibility. In a dual target RSVP participants must flexibly respond to both T1 and T2, a reduction in flexibility means that the detection of a second target is more difficult because the individual is so focused on the first target. Olivers and
Nieuwenhuis (2006) suggest that when participants are less focused on one source of information they can flexibly attend to multiple stimuli and therefore the AB is reduced. Experimental evidence to support this shows that the AB diminishes when participants share their attentional resources between the RSVP and another task, such as listening to music (Olivers & Nieuwenhuis, 2005) or when they are encouraged to adopt a more distributed state of attention (Olivers & Nieuwenhuis, 2006).
Olivers et al. (2007) have suggested that the AB is caused by an “overzealous application of control over the input” (pp. 137) and they account for the AB using the ‘overinvestment hypothesis’. Completing the RSVP task requires a balance between allowing for the detection of targets and inhibiting the selection of distracters; too
much emphasis on selecting targets means that some distracters may be processed, too much emphasis on inhibiting distracters may lead to some targets being missed. When T1 is detected the attentional set loosens and T1+1 may enter processing, if T1+1 is a distracter the set needs to tighten and correct itself, meaning that if T2 is presented at early lags following lag 1 it will be missed as more emphasis is placed on inhibition. Following from this it may be proposed that the more flexible, and less controlled the set is initially, the less likely this problem will occur; the individual will strike a more appropriate balance between selection and inhibition, and the control system will not need to continually readjust, so allowing better performance and a smaller AB. The finding that T1 accuracy is also better for nonblinkers supports this.
Although Olivers and colleagues do not support the TLC model (Di Lollo, Kawahara, Ghorashi, & Enns, 2005) and instead propose that their work is an
extension of the early limited-capacity theories, their hypothesis does suggest that the AB is caused by limitations in top-down processing and control. Indeed the majority of recent studies reporting evidence of individual differences suggest a role for top- down control, but Olivers et al. (2007) state that their findings provide support for resource depletion accounts of the AB, and they deny that the AB is purely due to issues with the top-down attentional set. Can individual differences in AB magnitude also be explained by the TLC model? The model states that the top-down attentional set requires constant feedback to keep it active. When T1 is being processed feedback signals can no longer be sent and the top-down system loses control. If T1+1 is a distracter it will trigger an exogenous attentional set, and the endogenous set will have to be reconfigured to regain control. This reconfiguration causes switch costs which result in the AB. It may be suggested that when more cognitive control is exerted the set will be less flexible because more emphasis will be placed on stabilising the set to
prevent exogenous capture. Reconfiguring the top-down set once the exogenous system has gained control will therefore lead to greater switch costs and a larger AB. At this stage it is difficult to do more than postulate on the possible impact of control over the AB, but the individual differences in performance found in the current work, and previous reports in the literature do imply that the cognitive control of attention does have an impact upon performance in a RSVP.
3.4.2 The influence of cognitive control on carry-over of attentional set
Experiment Three does not provide strong statistical evidence that the top- down attentional set from a dual target RSVP will persist to a single target RSVP, however trends found within the data do suggest that carry-over may be occurring. In addition, the individual differences found in the experiment may have had an impact upon the findings of carry-over. First, the amount of control given to the task may not actually influence the possibility of a participant using the same attentional set in two different blocks, but it may influence the perceived carry-over. The rationale for using the RSVP methodology is that in a dual target block performance follows a U-shaped function (the AB). If the set adopted to complete the dual target block is not
abandoned when the block ends it will also elicit the same pattern in a subsequent single target block. If an individual does not show an AB in the dual target block there is no way of determining whether the attentional set used to complete the task has persisted to the single target block, because performance will be no different to the no-set-priming group. As a result the evidence for persistence of attentional set will appear to be weak.
In addition to this the control of the top-down attentional set may also have a more direct influence over the persistence of this set. It has been predicted that an
increase in cognitive control results in a larger AB because it reduces cognitive flexibility. Regardless of which theory one chooses to support, most researchers state that increased control leads to a larger AB (Martens et al., 2006; Olivers &
Nieuwenhuis, 2005; 2006; Olivers et al., 2007; Taatgen et al., 2007). If the set
becomes less flexible and more resources are allocated to the task the costs associated with switching set will be greater. If the costs of switching set outweigh the benefits there will be the less motivation to change set when the task demands change (Leber & Egeth, 2006). It is therefore hypothesised that individuals who have greater control will be more likely to suffer from the persistence of attentional set.
It is important to clarify which level of cognitive control (micro or macro) the author is referring to at this point. Recall that macro-control is defined as the overall control over the goal representation, and the set established to complete the goal, whereas micro-control is defined as direct control over the task. Although macro- control is essential for the balance between flexibility and stability it is postulated that the above work is more related to the definition of micro-control; how much control (or effort) one puts into the task and the cognitive set. A high level of control (leading to a larger AB) will consolidate the set and a low level of control (smaller AB) will allow the set to be more passive and flexible. This means that increased micro-control will increase the chance that the set will carry-over.
3.4.3 Implications for previous findings
It is worth noting that the findings from Experiment Two may also have been influenced by individual differences in AB magnitude. In Experiment Two the comparison of the first block completed by each group only revealed an AB in the dual target block in the planned contrasts, not in the main effects (see page 62).
Additionally, participants showed a mean detriment in performance in a dual target block of 11.05%. When comparing this to the detriment found in the dual target block completed in Experiment Three (21.62%) it is clear to see a big difference in the size of the blink. Obviously the experiments differ in terms of the stimuli used, the design implemented, overall RSVP performance, and the cuing effect found in Experiment Two, yet this is still an important difference. It suggests that the participants in Experiment Two were clustered towards the nonblinker end of the scale, therefore reducing the size of the AB in a dual target block, and any carry-over in the following single target block.