Task Switching and Attention: an Eeg Study of Distraction by Food Stimuli

Task switching and attention: An EEG study of distraction by food stimuli

Student ID: 610018120

PSY3250

University of Exeter

Ackwnoledgements

I would like to thank my supervisor, Dr. Aureliu Lavric, for the invaluable guidance he provided throughout this project and for being a continuous source of inspiration. I would also like to thank my parents, whose encouragement and support made all of this possible.

Abstract

We investigated the largely neglected subject of distractibility during task switching using images of food as distractors. Participants were required to abstain from eating for 3h prior to the session, since hunger levels were previously correlated withattentional bias for food stimuli.They completed a computer task where they were required to switch between identifying colours and shapes. On half of the trials, task-irrelevant images (food- and office-related images) appeared on both sides of the relevant stimulus and subjects were asked to ignore them. We acquired measures ofperformance (RT and error rate), brain activity (EEG) and food-related motivation (food craving trait questionnaire and hunger measures during the session.Contrary to our expectations, although the presence of a distractor did affect performance in a negative way, participants were not more distracted on switch trials compared to repeat, as the distractor effect was not modulated by switching; behavioural and EEG data show similar distractibility on switch and repeat trials. Secondly, we found that motivationally salient stimuli (food image) are processed more than neutral (office) items, reflected in ERP lateralization- an effect that was not greater during task switches than during task repetitions. Lastly, we found evidence of a switch cost, reflected in RTs and error rates, but not in ERPs and contrary to expectations, this switch cost was not reduced by longer preparation times. This could be explained by the fact that participants were busy preparing for the eventual appearance of the distractor but more research will be needed to fully understand this effect.

Imagine the following scenario: You are driving on a busy motorway when the person sitting on the passenger seat asks you a question. You turn the radio down and answer. Meanwhile, you notice a billboard on the side of the road, advertising a new meal at your favourite restaurant so you decide to stop on the way home and buy some food. Situations like this are not uncommon and people often find themselves having to switch back and forth between various activities (Monsell, 2003). In this particular case, the tasks are demanding enough that they engage one’s cognitive functions but nevertheless, attention is still drawn to a distractor which is entirely irrelevant to both the phone conversation and driving. Although, there is considerable literature in the area of task-switching, not much is known about the interaction between task-switching and attention, and even less about the role that motivationally salient stimuli, such as food items might have in mediating this relationship (Engelmann & Pessoa, 2007). Thus, the present experiment investigated these two largely neglected areas, by employing a task-cuing paradigm with images of food and office items serving as distractors for participants who were in a hungry state.

Responding to daily demands in a flexible way requires adopting a task set (Monsell, 1996), a set of mental representations that allows people to identify relevant stimuli, select appropriate responses and act in accordance with contextual requirements (Logan & Gordon, 2001). In laboratory settings, regulation of these mechanisms of cognitive control has been studied using task-switching experiments, initially run by Jersild (1927). In a task-switching experiment, participants are usually presented with stimuli that afford two or more responses (Monsell, 1996) and are trained on tasks which usually involve categorization or identification (Vandierendonck, Liefooghe & Verbruggen, 2010). Subjects then perform the tasks which can either change from trial to trial or remain the same on successive trials (Kiesel, Wendt, Jost, Steinhauser & Falkenstein, 2010). Comparing the performance in these two conditions (Nieuvenhuis & Monsell, 2002) has shown that switching between tasks results in disrupted performance (Rogers & Monsell, 1995), expressed through significantly longer response times (RT) and higher error rates on switch trials compared to repeat trials (Meiran, 1996).

There are various methods in which the participant can be informed about which task needs to be performed, but the most commonly used procedure is the task-cuing-paradigm (Monsell & Mizon, 2006), where the required task is specified by an explicit cue which appears before or with the stimulus (Altman, 2004). This paradigm makes it possible for the experimenters to manipulate preparation times by altering the interval between cue and stimulus (CSI- cue-stimulus-interval; Meiran, 1996). Advance knowledge of the upcoming task in the form of longer CSI has been linked to decreased switch costs, termed the ‘RISC’ effect (reduction in switch cost with increased preparation time; Longman, Lavric & Monsell, 2013). However, even with CSI as long as 1420ms (Longman, Lavric, Munteanu & Monsell, in press), a complete elimination of the switch cost does not seem to be possible and two competing explanations have been offered for the source of this asymptotic ‘residual’ switch cost: Task-Set Reconfiguration and Task-Set Inertia.

Firstly, cognitive control theorists explain switch cost though Task-Set Reconfiguration (TSR; Yeung & Monsell,2003), which is the need to prepare a set of endogenous cognitive processes, such as retrieving goal states, during the CSI (Nieuwenhuis & Monsell, 2002). Empirical evidence for TSR was provided by Rogers and Monsell (1995), who employed a task-switch categorization paradigm and varied the time participants had to prepare before the task. Their results showed that completing TSR in advance reduces error rates and RTs, supporting the RISC effect. An advantage of this model is that it accounts for the presence of residual switch costs. According to Rogers and Monsell (1995), irrespective of the length of CSI, part of TSR can only be done after stimulus onset. A more recent theory (De Jong, 2000), considers the residual switch cost as an average of trials where TSR was successfully completed before the stimulus (no residual switch cost) and trials where TSR “fails to engage” before the stimulus.

An alternative explanation of switch cost is provided by the carryover of stimulus-response mappings of the previously relevant task, termed Task-Set-Inertia (TSI, Allport, Styles & Hsich, 1994). Recently, the term “attentional inertia” has been put forward (Longman et al., 2013), to suggest that when the participant is cued to perform a task, the previous task-set continues to interfere and thus switch cost arises because participants’ attention is drawn to properties of the stimuli that were previously relevant (De Jong, 2000). The majority of distraction studies so far have focused on dual-task interference (Strayer & Drews, 2007) and thus, there is a gap in research investigating the interaction between attention and task switching. Among the few studies conducted in this area, which support the idea of attentional inertia, is a study by Longman et al. (2013). In this study, participants had to respond using the keyboard to pictures representing one of four faces that had one of four letters superimposed on the foreheard and each task was associated with a location on the screen. Task-cuing was employed using an auditory cue and participants had either 200ms or 800ms to prepare before stimulus presentation. Eye-tracking analysis showed that switching tasks determined participants to delay their attention to task-relevant attributes and wrongly allocate attention to the previously relevant location. A more recent study (Longman, Lavric, Munteanu, & Monsell, in press), used the simultaneous presentation of three digits associated with a location and a classification task to isolate attentional inertia from general distractibility. Using eye-tracking, this study showed that people have a tendency to wrongly allocate attention on the location that was relevant on the previous trial, and not on other irrelevant locations. This effect was reduced by extending the CSI to over one second, providing support for the contribution of attentional inertia to ‘residual’ switch cost. Conflicting evidence comes from Lien, Ruthruff and Johnson(2010), who claimed that the contribution of attentional processes to residual switch cost is not as significant as previously thought. Their experiment used a contingent capture paradigm with uninformative cues, which either had the same colour and location as the target stimulus or different ones. Results showed that these cues only affected performance when the colour was relevant, with no significant evidence suggesting attentional capture by stimuli whose colour was previously, but no longer relevant (i.e. on switch trials).

In these cases, distraction was represented by a stimulus thatrecently required a response, and thus, was relevant for the allocation of attentional resources. However, in daily life, people often get distracted by stimuli which are entirely irrelevant to the task at hand and that should theoretically be ignored (Forster & Lavie, 2008).Considering that not all perceived information is relevant, selective attention is needed so that people can focus on stimuli which are needed for the task at hand, while keeping less relevant information in the background (Lawo, Fels, Oberem & Koch, 2014). Understanding these failures of selective attention during task switching is a largely neglected area of study.In fact, a few experiments have shown that the presence of irrelevant distractors during a single-task can affect performance to the same extent as response-competing distractors (Theuuves, 1991). As a task switch possibly resets attention and fully engages cognitive abilities (Lavie, 2010), it would be interesting to gain a better understanding of the effect of irrelevant distractors in this setting, so that ways to avoid their interfering effects can be identified (Forster & Lavie, 2008). A good theoretical background is provided by the ‘Load Theory’ (Lavie, de Fockert & Viding, 2004) which suggest that attentional control is worse under high levels of cognitive load, and thus, it would be expected that people are more easily distracted by irrelevant stimuli during switches. Another question logically follows: if attention is, in fact ‘grabbed’ by external irrelevant events, is there any difference in the type of distractors that are presented?

The second theme of this paper will attempt to answer this latter question by focusing on a specific type of distractor – stimuli that have motivational significance for the participant, as they seem to be more effective in affecting task-relevant responses (Vergoeven et al., 2010). For example, pictures depicting food items have been shown to increase attentional bias for food related-stimuli (Mogg, Bradley, Hyare & Lee, 1998) and were associated with enlarged positive ERP potentials over posterior sites (Stockburger, Weike, Hamm & Schupp, 2008), with even stronger effects when participants were in a hungry, rather than satiated state (Engelmann & Pessoa, 2007). Although laboratory based, this method of measuring distraction has a high level of external validity, as in daily life, distractors are usually more attractive than task-relevant stimuli (Forster & Lavie, 2008)

Studying the effect of distracting food stimuli can have major implications for developing interventions in treating obesity, as reactivity to food cues is potentially modifiable (Castellanos et al., 2009). Even though genetics are crucial in the development of this issue, this dramatical rise in the prevalence of obesity can only be explained by the involvement of a different factor, in the form of environmental influences (Peters, Wyatt, Donahoo & Hill, 2002).The incentive sensitization model of obesity (Nijs & Franken, 2012) hypothesizes that because of the abundance of food cues in the environment, people become sensitized to food stimuli, as their attentional processing is enhanced and automaticised (Nijs, Franken & Murris, 2010) and this, in turn, contributes to excessive food intake (Castellanos et al., 2009). Evidence for this model was provided by a longitudinal study conducted by Calitri et al (2010), which showed that weight gain was positively correlated with earlier attentional bias for food and was successfully predicted 1 year in advance.

The current project aims to link the areas of task-switching, attention and motivation by using a task-cuing paradigmclosely modelled on that of Lavric, Mizon and Monsell(2008) with images of food items as lateralized distractors. Our first hypothesis, based on task-switching research, predicts that task switches will be associated with greater distractibility than task repeats, as switches would put greater load on WM, which, based on Lavie’s ‘load theory’ of attention (Lavie, de Fockert & Viding, 2004), predicts larger vulnerability to distractors.Secondly, we hypothesize that food, as motivationally salient stimuli, will be processed more and affect performance on greater extent compared to neutral office items (larger ERPs) and that this difference will be larger on switch trials. Lastly, based on previous findings in task-switching literature, we predict that performance on the colour-shape task will be affected more on switch trials compared to repeat trials (slower RT, larger error rates and larger ERPs) and that this switch cost will be reduced by longer CSI.

Methods

Participants

The participants were 23 undergraduate students at the University of Exeter (11 men and 12 women), aged between 19 and 24 (M=20.82, SD=.936). Participants were a convenience sample recruited personally by the experimenters. In exchange for their time, they were compensated with up to £5 (a flat rate of £2 plus performance related bonus based on minimizing RT and error rates). All participants gave informed consent and the study was approved by the University of Exeter School of Psychology ethics committee. One participant was excluded from ERP analysis due to large number of artifacts.

Design

The experiment used a within subjects design with task-relevant stimuli (coloured shapes) presented centrally and irrelevant images of food or motivationally-neutral office items presented concomitently to the left or right. Measures of Reaction Times and Error Rates were taken alongside measures of preferential processing of motivationally-salient distractor in the form of posterior lateralized ERPs (visual processing area). The independent variables weretransition (switch vs repeat), distractor (present vs absent), response congruence (congruent vs incongruent) and CSI (short vs long).

Participants were approached by the experimenters and asked to take part in a study investigating the effects of task switching on performance. They were informed that they will be required to restrain from eating for 4 hours prior to the experimental session. Before the beginning of the session, they were given consent forms, giving a brief overview of the study and specifying that the information is anonymous and that they have the right to withdraw at any point (Appendix A).

During the EEG set-up, participants were asked to complete two questionnaires measuring hunger levels and food-craving trait (Appendix B). Hunger levels were assessed before and after the session with a short 3-item questionnaire (e.g. “How strong is your desire to eat”) and food-craving was assessed once on a 6-point scale ranging from 1=never to 6=always (e.g. “I feel like I have food on my mind all the time”).

The study used a task-cuing paradigm, presented using the software E-Prime, with participants being required to switch between identifying colours and shapes, while ignoring irrelevant pictures (food and office items) which appeared bilaterally at unpredictable times on half of the trials. The task changed with a probability of 0.33 and the cue-stimulus interval (CSI) was either long or short and distractors appeared either relative to cue onset (long CSI: 240 ms, 360 ms, 480 ms, 600 ms and short CSI: 40 ms, 160 ms, 280 ms, 400 ms) or relative to stimulus onset (40 ms, 160 ms, 280 ms, 400 ms). There was a filler trial at the start of each block in order to create a transition (switch or repeat) for the first analyzed trial. No distractors were presented on these and filler trials were not considered for analysis.Before starting the main experiment, participants completed three practice blocks. First, they practiced the colour-task and shape-task separately, then did a practice block on switching between the two and finally did a shorter block where they got familiarized with the main procedure, which involved distracting pictures.

The main session, which lasted approximately 50 minutes, consisted of a total of 1164 trials, divided in 12 blocks with 96 trials each, plus a start up filler trial before each block.

Information appearing on the monitor instructed participants to keep their gaze on a fixation cross located in the middle of the screen. On each trial, a cue preceding the onset of the stimulus indicated whether they should attend to the shape (SHAPE or FORM) or colour of the stimulus

(COLOUR or PAINT) (See Figure 1).