Individual but not fragile: Individual differences in task control predict Stroop facilitation

Abstract

The Stroop effect is composed of interference and facilitation effects. The facilitation is less stable and thus many times is referred to as a “fragile effect”. Here we suggest the facilitation effect is highly vulnerable to individual differences in control over the task conflict (between relevant color naming and irrelevant word reading in the Stroop task). We replicated previous findings of a significant correlation between stop-signal reaction time (SSRT) and Stroop interference, and also found a significant correlation between SSRT and the Stroop facilitation effect—participants with low inhibitory control (i.e., long SSRT) had no facilitation effect or even a reversed one. These results shed new light on the “fragile” facilitation effect and highlight the necessity of awareness of task conflict, especially in the Stroop task.

Introduction

Executive control is a key human function that mediates the ability to maintain goal-directed behavior (Banich, 2009, Miller and Cohen, 2001, Miyake et al., 2000, Shallice and Norman, 1986). A popular task to investigate this process in the lab is the Stroop task (Stroop, 1935). In the common task, individuals are asked to identify the ink color in which a word is printed (i.e., word color) while ignoring the word meaning. The word color and word meaning can be either congruent (e.g., RED written in red), incongruent (e.g., GREEN written in red) or neutral (e.g., XXXX written in red). It is commonly assumed that incongruent trials create a conflict, and a control mechanism is recruited to settle this conflict. Hence, the reaction time (RT) for incongruent trials is commonly longer than RT for neutral trials. This is known as the interference effect, which is a very strong and stable effect (MacLeod, 1991). In congruent trials, both word color and word meaning activate the same response; hence, RT is usually faster than in neutral trials. This is known as the facilitation effect, which is a much smaller and less stable effect than the interference effect (e.g., Dalrymple-Alford & Budayr, 1966). The facilitation effect is sometimes absent (Kalanthroff, Goldfarb, and Henik, 2012, Mathis et al., 2009), and in many studies referred to as a “fragile effect” (e.g., Fuentes and Ortells, 1993, Goldfarb and Henik, 2007, Logan and Zbrodoff, 1998, MacLeod and MacDonald, 2000). The current study aims to address the fragility of the facilitation effect and examines whether individual differences in the ability to execute control produce differences in the size of the facilitation effect.

Several neuroimaging studies demonstrated that the anterior cingulated cortex (ACC)—a brain area thought to monitor conflict (Botvinick, Braver, Barch, Carter, & Cohen, 2001; Botvinick et al., 1999, Carter et al., 1999, Carter et al., 1998)—is more activated in the incongruent condition than in the neutral condition (e.g., Bench et al., 1993; Carter, Mintun, & Cohen, 1995; Milham et al., 2002, for older participants). Interestingly, however, these studies also showed that not only the incongruent Stroop stimuli triggered higher ACC activations than the neutrals did, but the congruent Stroop trials did so too. Hence, the findings of ACC activation mentioned above imply that conflict occurs not only in the incongruent condition but also in the congruent condition.

The contradiction between the behavioral (RT to congruent trials faster than to neutral trials) and neurological (higher ACC activation to congruent than neutral trials) findings has been subject to a number of recent investigations. Goldfarb and Henik (2007) suggested that the contradiction can be settled by distinguishing between two conflicts reflected in the Stroop task—informational (between the incongruent word and ink color) and task (between relevant color naming and irrelevant word reading). While the information conflict involves the content of the stimulus and the response needed, differing between congruent and incongruent Stroop stimuli, task conflict involves the task associated with the stimulus, and differs between congruent and non-word neutral Stroop stimuli (Kalanthroff, Goldfarb, & Henik, 2012). This is consistent with the suggestion that a stimulus has the ability to evoke performance of a task that has a strong association with it (Allport and Wylie, 2000, Rogers and Monsell, 1995, Waszak et al., 2003) and specifically, that words trigger an automatic tendency to read written words (MacLeod & MacDonald, 2000). Monsell (2003) proposed that a cognitive task, and the efficiency with which we perform it, results from an interplay between 2 mechanisms: (1) endogenous control, which emphasizes the role of deliberate intentions governed by goals, and (2) exogenous influences, which are caused by alternative possible tasks available by the stimulus. In the Stroop task, endogenous control will lead to focusing attention on the color naming task, whereas word reading will be the strongest exogenous influence. Monsell’s proposal seems to fit with the task conflict reasoning. Recently, David et al. (2011) found evidence, in an event-related potential (ERP) study, for the existence of task conflict in the Stroop matching task; Steinhauser and Hubner (2009) found evidence for the dissociation of task conflict from the response conflict in the ex-Gaussian distribution analysis.

Wilkinson and Halligan (2004) suggested that one of the cases in which behavior expected from brain activation will not appear is when another brain structure contributes to the reduction of this behavior. Accordingly, the contradiction between behavior and brain activation in the Stroop task indicates the existence of an active control mechanism aimed at eliminating the task conflict. In line with this, Goldfarb and Henik (2007) claimed that the task conflict is usually not seen due to a very efficient task control mechanism. Thus, task conflict does delay responding to the congruent condition but (due to the efficient control mechanism) not enough to make it slower than responding to the neutral condition. Task control is a cognitive mechanism that is responsible for solving the task conflict or the ‘task set competition’; namely, the need to perform a target task and play down an irrelevant task (Banich, 2009, Goldfarb and Henik, 2007, Monsell, 2003, Monsell et al., 2001). Recently, Braver and colleagues (Braver, 2012, De Pisapia and Braver, 2006) suggested the dual mechanism theory of pro-active (reflecting the strategic block-wise control) and reactive (reflecting the stimulus driven trial-by-trial control) mechanisms. Though this framework does not account for task conflict, it seems reasonable to assume that task conflict will usually become visible when pro-active control (or the ‘task demand’ component introduced in earlier models, e.g., Botvinick et al., 2001) is low. Goldfarb and Henik (2007) tried to ‘put the task conflict (pro-active) guard to sleep’ by increasing the proportion of nonword neutral trials and by providing cues for whether the upcoming trial was going to be a neutral or a conflict one (i.e., congruent or incongruent). This reduced the task conflict control and resulted in a reversed facilitation effect in which the congruent RT was longer than the neutral RT—a behavioral evidence for the task conflict. In a more recent study (Kalanthroff, Goldfarb, Usher, & Henik, 2012), we found that the reverse facilitation effect appears in a similar procedure even if no incongruent trials exist. This suggests that task conflict appears whenever there is an alternative possible task regardless of the response it initiates. Thus, it seems reasonable to assume that RTs to Stroop congruent trials are contingent upon the efficiency of the task control mechanism.

Task conflict and the idea that the direction of the facilitation effect depends on the activation of an irrelevant task has been discussed in the task-switching literature (e.g., Aarts et al., 2009, Braverman and Meiran, 2010, Rogers and Monsell, 1995). Considering that task switching reduces pro-active control, it is not surprising that a reverse facilitation is often seen in task-switching studies (see also Aron, Monsell, Sahakian, & Robbins, 2004).

The current study aims to investigate whether individual differences in task control can explain the variance in the Stroop facilitation effect. As mentioned earlier, the Stroop facilitation effect is small, unstable and often referred to as “fragile”—some individuals and some subject group samples do not show the effect (e.g., Fuentes and Ortells, 1993, Kalanthroff, Goldfarb, and Henik, 2012, Logan and Zbrodoff, 1998, MacLeod and MacDonald, 2000). We predict that individual differences in the efficiency of the task control mechanism can explain why some individuals have a facilitation effect and some do not.

In the Stroop task subjects have to inhibit their irrelevant but automatic, prepotent reading responses (Miyake et al., 2000). Such inhibition is needed to improve performance in both incongruent and congruent trials. La Heij and Boelens (2011) recently suggested that children’s low ability to manage task conflict can result from their immature inhibition of a prepotent response mechanism. Accordingly, we believe that the ability to inhibit a prepotent response can predict the magnitude of both the Stroop interference and facilitation effects.

A task that is commonly used to measure inhibition of a prepotent response is the stop-signal task (Logan, 1994, Logan and Cowan, 1984), which examines the ability to suppress an already initiated action that is no longer appropriate. A great deal of research has been carried out in the past few decades with the stop-signal task (for review see Verbruggen & Logan, 2008). In the classic task, participants are asked to address a visual stimulus (go signal) with a motor response as fast as possible. In about 25% of the trials, an auditory stimulus (stop signal) comes right after the visual go signal. Participants are instructed to inhibit their motor response when they hear the stop signal. The duration between the go signal and the stop signal (SSD; stop-signal delay) is subjected to a tracking procedure—in the subsequent stopping trial, the SSD will become a bit longer (more difficult) following successful inhibition, or a bit shorter (easier) following erroneous response to the stop signal. Eventually, it becomes possible to estimate the stop-signal reaction time (SSRT), which is the time needed for successful inhibition. Logan and Cowan, 1984, Logan et al., 1984 compared the performance in the stop-signal task to a ‘horse race’ between the go process (triggered by the go signal) and the stop process (triggered by the stop signal). While RT is the time needed for the go process to finish, SSRT is time needed for the stop process to finish. Logan et al. (1984) argued that “response inhibition phenomena are consistent with a hierarchical theory of attention in which a high level process determines the significance of incoming stimuli and decides whether to abort the current stream of thought and action…” (p. 290). Indeed, the stop process represents high-level control and SSRT has proven to be an important measure of cognitive control (Verbruggen & Logan, 2008), and has proven useful in developmental and pathology studies (for reviews see Boucher et al., 2007, Logan, 1994).

In a previous study (Kalanthroff, Goldfarb, & Henik, 2012), we analyzed stop-signal control failure trials (i.e., erroneous response to stop-signal trials). In that study, we found longer RTs in the Stroop congruent condition compared with the Stroop neutral condition. Thus, a momentary failure in stopping results in both an erroneous response to a stop signal and in slow management of the task conflict. These results strengthened the notion that there could be some overlap between the inhibition of a prepotent response and task control or the control needed to manage the task conflict in the Stroop congruent condition. This fits in with Friedman and Miyake’s (2004) suggestion that stopping and Stroop belong to the same latent variable.¹

Here we suggest that variance in the Stroop facilitation effect is not random and that it can be explained in terms of individual differences in task control. The current study aims to disclose a factor that contributes to the variance in the Stroop facilitation effect—the task conflict that relies on inhibitory control. According to the notion that task control relies on inhibition of a pre-potent response, we predicted that individual differences in SSRT would correlate with individual differences in Stroop facilitation. More specifically, we expected that a long SSRT (which indicates low inhibitory control) would predict no (or even reversed) Stroop facilitation, and that a short SSRT (high inhibitory control) would predict large facilitation.