Working memory is an essential cognitive system in which incoming information is maintained and processed. The importance of working memory is that these processes occur despite interference or distraction (Miyake and Shah 1999
). There are many tasks that purportedly measure working memory, but a common paradigm used by researchers is the n
-back task. In a typical n
-back task, participants are presented with a series of visual or verbal stimuli. For each stimulus presentation, they are asked whether the current one matches the stimulus previously seen n
trials before. For example, in a 3-back letter task, participants have to decide whether the presented letter (e.g., “c”) is the same one presented in trial number n
-3. The n
-back task is considered a working memory task because it requires constant encoding of information, a temporary store for each stimulus presented, and a continuous processing or updating of each subsequent presentation. At the same time, the participant must inhibit irrelevant items, which must be removed from the memory store. Because of the cognitive subprocesses, the n
-back task is also described to be measuring executive attention (Kane and Engle 2002
). However, unlike a traditional letter-target n
-back task, we embedded emotional and non-emotional distractors that flanked the centrally located target probes. The specifications for these distractors were varied across three studies so as to examine the contributions of ability EI to two key aspects of executive functions: emotional versus non-emotional inhibition and working memory updating.
1.1. Emotional Intelligence Abilities
A conceptual debate on the basis of EI has been ongoing for decades with several competing and complementary models being developed. The theoretical development of EI can be divided between two main perspectives: (1) trait EI and (2) ability EI (Mayer et al. 2008
). From the trait perspective, EI is about one’s self-perceptions and dispositions. Such data are usually obtained via self-report measures. The commonality between the trait perspective and the self-report perspective is that both measure subjective perceptions of one’s emotional abilities. They typically involve self-report of emotional experiences and people’s behavioural preferences or styles relating to emotional expressions. A key differentiator of the trait EI perspective is that such measures do not correspond to the traditional paradigm for assessing general intelligence (g
) or specific problem-solving abilities (e.g., fluid reasoning).
More important to the present set of studies is the ability perspective of EI (Mayer et al. 2016
). The ability EI perspective examines those capabilities specialized for the processing of emotional information. The main method of investigation are performance assessments of emotional-based competencies and abilities in which objective answers or optimal responses exist, as determined by an expert or general consensus. The Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT; Mayer et al. 2002
) is a prototypical test of ability EI. The ability model posits that EI is best viewed as an intelligence that is based on (or built upon) the abilities to use or reason about emotions. Mayer et al.
) suggests that ability EI focuses on hot information processing. Several researchers, notably Gutiérrez-Cobo et al.
), have found correlations between participants’ performance on emotion-related tasks with ability EI but not on trait EI measures. In other words, EI consists of a class of abilities that involve processing emotional information and the effective utilizations of emotional information towards fulfilling the goal at hand.
If ability EI pertains in the way just described, perhaps a logical step forward in its conceptualization is to examine its relationships and broader contributions to human cognition. Mayer et al.
) as well as Elfenbein and MacCann
) suggest that there are two key facets of ability EI: Experiential EI and Strategic EI. According to Mayer et al.
), the facet of Experiential EI essentially involves the abilities to identify or perceive the emotions of others. It is also considered to be reflective of lower-order abilities in emotional processing. On the other hand, Strategic EI describes the online emotional reasoning abilities and how emotions can be appropriately managed to suit the context at hand. It is also considered to be indicative of higher-level or conscious emotional information processing. Experiential EI and Strategic EI together form the binary aspects of EI as described by Mayer et al.
). Abilities under the Experiential EI facet are used for rapid perceptual processing of emotional information while Strategic EI abilities are used for deliberate (or strategic) reasoning and control of emotional information (Elfenbein and MacCann 2017
). Importantly, Strategic EI involves the abilities to modulate or regulate emotions in oneself and manipulate others in the pursuit of one’s goals (Mayer and Salovey 1997
). The two facets of Experiential and Strategic are central to the Mayer et al.
) conceptualization of ability EI, such that the MSCEIT provides two area scores that correspond to them, respectively. For example, the Experiential EI score is obtained from combining Branch 1: Perceive Emotions and Branch 2: Facilitating Thought whereas the Strategic EI score is obtained from combining Branch 3: Understanding Emotions and Branch 4: Managing Emotions.
A recent study by Gutiérrez-Cobo et al.
) examined the relationships between EI and hot/cold executive functions (i.e., cognitive control). In their study, participants completed measures of EI according to either the self-report ability model (Trait Meta-Mood Scale, TMMS), self-report mixed model (Emotional Quotient Inventory—Short Form, EQi:S), or performance-based ability model (MSCEIT). The researchers also had participants complete both hot and cold variants of an executive function task (i.e., go/no-go task). For the hot variant, participants had to press a button (“go” trials) for a target emotional face and to refrain (“no-go” trials) for a neutral face, and vice versa in different blocks. For the cold variant, participants completed a go/no-go task using shapes and colours as stimuli. The researchers found that, overall, positive correlations (in the order of r
= 0.15 to r
= 0.18) between participants’ scores on the managing branch on the MSCEIT and their reaction times when “go” stimulus was an emotional face but not with neutral faces. Scores on the MSCEIT’s managing emotion branch were also negatively associated with false alarm rates for happy and angry faces (in the order of r
= −0.15 to r
= −0.19). While these associations were small, these findings suggest that Strategic EI abilities, which entail the managing branch, could contribute positively to one’s executive control for emotional conflict.
In a follow-up study, Gutiérrez-Cobo et al.
) examined whether EI was linked to working memory capacity (WMC) and whether this relationship was different if the WMC task was of a hot (affective) or cold variant. They had participants complete a hot n
-back task using angry, happy, and neutral face stimuli (e.g., “Was this happy face the same happy face n
-steps ago?”), as well as to complete a cold n
-back task (e.g., whether the presented letter is the same as the one n
-steps ago). Once again, they found that the managing branch of EI was positively associated with performance on the hot (or emotional face) n
-back task but not on the cold variant (i.e., updating letters). This finding was intriguing and held conceptual implications to the broader emotional regulation literature, as previous researchers (e.g., Schmeichel and Demaree 2010
) found that WMC could be related to one’s ability to regulate emotions. In Gutiérrez-Cobo’s et al. study, WMC was implicated in the managing branch of EI. This suggested that the ability to regulate emotions both in self and in others (the definition of the Strategic EI facet on the MSCEIT) requires the cognitive control of emotional information. These findings are also consistent with other studies that demonstrated emotional working memory training could improve emotion regulation (Schweizer et al. 2013
). In other words, Strategic EI abilities are utilized when the cognitive task requires a degree of emotional control and inhibition.
One important limitation of Gutiérrez-Cobo et al.
) is that their studies did not specify the locus of the EI contributions to the performance on the cognitive tasks. In their study using a hot and cold n
-back task (Gutiérrez-Cobo et al. 2017b
), the targets were either emotional faces (hot variant) or consonant letters (cold variant). As such, it was uncertain whether the implicated EI processes were in relation to: (1) affective processing (i.e., simple updating of emotional information) or (2) affective inhibition (i.e., the ability to inhibit an emotional context when performing on the task). Another minor limitation is that Gutiérrez-Cobo et al.
) did not employ a more standardized way of calculating n
-back performance. For example, they calculated n
-back accuracy based on the total number of hits instead of the more commonly used indices of percentage (or proportion) of hits divided by the number of trials (i.e., as used in signal detection theory; Jacola et al. 2014
; Kane et al. 2007
). They also calculated “miss rate” as the total number of “no” responses to n
-back targets. Additionally, Gutiérrez-Cobo et al.
) used a two-key response version of the n
-back task, but they did not take the rate of correct rejections (or false alarms) into consideration. Correct rejection rates are a useful metric for calculating n
-back accuracy (Jacola et al. 2014
). Taken together, it was not possible to determine whether the contributions of ability EI were due to the challenges in inhibiting responses to a non-target probe or due to failures when responding to a target. In our studies, we also employed a two-key administration to ensure separation of participants’ responses but also utilized participants’ hit rates and correct rejection rates to obtain a better estimate for emotional n
Another limitation of Gutiérrez-Cobo’s et al.
) study is that they did not find significant correlations between response latencies for correct (or hit) trials on the n
-back task with any of their EI measures, including the MSCEIT. We find this non-relationship to be surprising because performance on the n
-back task is not solely restricted to accuracy rates. Importantly, response latencies have previously been found to predict unique individual differences in relation to intelligence. For example, Hockey and Geffen
) found stable and positive correlations between cognitive abilities and correct response times on an n
-back task. Similarly, Gevins and Smith
) found that participants in a high cognitive ability group (as measured using the WAIS-R) were significantly faster on the n
-back task when compared with low-ability participants.
In part to answer these limitations, as well as to explore the relationships between EI and working memory in greater detail, the n
-back task used in the current study held the type of target stimuli constant (letters only) while varying the affective context demands of the task. We created a task called the Emotional Flanker N-back (EFNB), which was modified from the original Emotional Face N-back task (EFNBACK; Ladouceur et al. 2009
). In the original EFNBACK, the task consisted of presenting a standard letter n
-back paradigm wherein each trial was either flanked by emotional faces (happy, fearful, neutral) or no picture. In our EFNB task, we kept the happy and fearful faces but changed the neutral/no-picture conditions. Instead of using neutral faces or no faces for the cold task variant, we used: (1) geometric shapes and (2) letters for the flanking distractors. The purpose of using shape flankers and letter flankers was to determine whether there was a difference when responding in the presence of affective or non-affective flankers within the same experimental paradigm and whether this difference was related to participants’ EI abilities.
Another difference from the original EFNBACK task was that we included a third memory-load condition (Study 1). In the original task, there were only two memory-load conditions (0-back, 2-back). There were two reasons for adding a third memory load (3-back) into the EFNB task. First, for any increase in N
, there is an increasing demand in processing load. As load increases, participants have to shift or expand the attentional focus between the presented stimuli for comparison in working memory (McElree 2001
) as well as discriminate and inhibit stimuli that no longer need to be maintained in working memory (Kane et al. 2007
). In the context of our current study, we hypothesized that EI abilities would be implicated when inhibiting distracting emotional information and that this would be more readily observed as processing load increased. Second, EI-related differences might be more pronounced in the hot n
-back task as the demands on WMC increased. If EI is considered to be a measurable mental ability (Mayer and Salovey 1997
), then increasing the demands of an EI-implicated task should lead to further differences in participants’ performances on an emotional n
-back task. Such a proposal would also be congruent with MacCann et al.
), who argued for EI as a broad intelligence that could possibly impact cognitive performance. It would also be congruent with Mayer’s et al.
) updates to their original ability EI model; specifically, EI is a class of broad intelligence that is utilized when working on “hot” (or emotional) information (p. 3). Finally, the 3-back task also requires a considerable amount of mental effort and is cognitively challenging at this difficulty level (Jaeggi et al. 2010
). A person with higher EI abilities may be much better in managing the affective distractors and memory load demands as compared to a person with lower EI abilities.
Finally, we used emotional faces as flanking distractors for two reasons. First, faces are meaningful sources of information that receive preferential processing and bias the allocation of one’s attention to themselves (Eimer and Holmes 2002
; Fenske and Eastwood 2003
). Additionally, emotional stimuli—especially negative ones—are likely to be processed automatically and rapidly by the brain (i.e., through the amygdala, e.g., Diano et al. 2016
). Emotional face flankers capture one’s limited resources for information processing, thereby impairing performances on tasks measuring working memory (Dolcos et al. 2008
). Second, the use of face flankers allows for the comparisons of affective states in their impact on memory updating (e.g., between happy faces and fearful faces flankers). In the EFNB task, the faces flankers are to be ignored while working on the n
-back task. However, negative faces capture attention nonetheless, even when people are explicitly instructed to ignore them (Eastwood et al. 2003
). As such, negative face flankers are likely to disrupt n
-back performance more so than positive face flankers and by extension, the letter or shape flankers. However, higher EI individuals are expected to be less susceptible to this disruption because they are able to allocate more resources towards inhibiting emotional distractors (Checa and Fernández-Berrocal 2015
1.2. Purpose of Study 1
Study 1 had two main objectives. First, it aimed to explore the notion that EI abilities are differentially related to performance in a hot and cold working memory updating task. To this end, we utilized flanker distractors that were either affective (happy faces or fearful faces) or non-affective (letters and shapes) and embedded them in a standard letter n-back task. Second, we were interested to examine whether varying the memory load (0-back, 2-back, and 3-back) would have an impact on participants’ performances on the hot/cold variants and whether these were influenced by their EI. The 0-back load condition was akin to a monitoring task (e.g., go/no-go) and primarily taps inhibition control, while the 2-back and 3-back load conditions tap both inhibition and updating processes.
There were two sets of core hypotheses of interest: four relating to the EFNB task and a set of moderation hypotheses related to the EI abilities. First, we hypothesized that an increase in updating load would impede n
-back performance (i.e., reduced accuracy rates and increased reaction times) in both affective and non-affective flankers, thereby replicating the standard n
-back memory load effect. Second, we expected that performance would be worse for the affective flankers because emotional faces capture attention more readily than non-facial features (Eastwood et al. 2003
), thereby impeding n
-back performances in the affective flanker conditions even further vis-à-vis neutral flanker conditions. In other words, we hypothesized that n
-back performance would be significantly poorer for both the happy face and fearful face flankers conditions when compared to neutral flankers (letters and shapes). A third and related hypothesis was that n
-back performance would be poorer for the fearful flankers condition as compared to the happy flankers condition because negative emotions are thought to be more attention-grabbing than positive ones (Hodsoll et al. 2011
). Finally, we expected that n
-back performance would be least impacted by shape flankers when compared to letter flankers because the probes for our n
-back task were letters. As such, letter flankers were expected to generate more interference because they shared features with the probes (Baylis and Driver 1992
For specific EI-related hypotheses, we first hypothesized that n
-back performance on the hot 2-back and 3-back conditions would be significantly moderated by participants’ Strategic EI score on the MESCIT, as suggested by the positive correlations found in Gutiérrez-Cobo’s et al.
) study. We also expected that EI abilities would not be significantly related to n
-back performance when the flanking distractors were either neutral shapes or letters.
In summary, we manipulated the n
-back memory load and the valence of flanking distractors to critically test three interference effects: (1) the interference of affective faces on working memory performance as compared to neutral flankers; (2) the interference of happy face flankers versus fearful face flankers; and (3) the contributions (moderation) of Experiential EI ability and Strategic EI ability with specific regard to these interferences. The affective face distractors would reduce available cognitive resources allocated to working memory, but this impairment was expected to be less pronounced in participants with higher Strategic EI abilities. If EI was indeed an ability to cope with one’s own emotions in an adaptive manner, it would mostly likely have been implicated only when the cognitive task required inhibition of emotional information. This was in line with Mayer and Salovey’s
) notion that EI abilities are about accurately facilitating and inhibiting emotional signals in response to a situation. In our n
-back task, the focus was evidently more on the inhibition of emotional interference arising from the signals generated from emotional face distractors.