Working Memory and the Binding Approach

The description of WMC has obvious overlaps with the executive function literature. In fact, Friedman and Miyake (2017) recognise that their work on execu-tive function emerged from the same tradition as the working memory literature.

Moreover, the distinction between the common executive function factor, executive

attention, and working memory capacity are hard to distinguish. According to Wil-helm, Hildebrand, and Oberauer (2013), the primary function of WMC is for build-ing, maintaining and updating temporary, arbitrary bindings. Bindings are formed be-tween items and their characteristics, such as ordinal position, for their later recall in span tasks. Therefore, complex span tasks such as OSPAN could be an example of the binding approach, as the task requires continuous updating of temporary bindings between each word that has to be remembered and its position within the list (Wil-helm et al., 2013). On one trial, new temporary bindings between items and positions have to be created, and maintained throughout processing. On the next trial, those old bindings must be updated through the creation of new temporary bindings between items and position markers. From this perspective, WMC is determined through in-terference among bindings. However, bindings are arguably employed under simple span tasks as well.

Unlike working memory tasks, short-term memory or simple span tasks do not require participants to manipulate information. Task items are simply stored for a short amount of time before being recalled. Short-term memory is the result from synchronized flexible recruitment of language, attentional, and serial order pro-cessing systems. To explain propro-cessing under complex span tasks, Unsworth and Engle (2007) put forth the dual-component theory, which proposes an active mainte-nance component, which they term primary memory (PM), and a controlled cue-de-pendent search and retrieval component, which they refer to as secondary memory (SM; Unsworth & Engle, 2007). Under the dual-component theory, items are sent from PM to SM when new or distractor items are presented which require processing (Unsworth & Engle, 2007). Items are then pulled from SM for recall. The reliance on components is dependent upon what the task requires of processing; simple tasks typically require PM more than SM, as less processing is required for the items to be shifted into SM than complex span tasks (Unsworth & Engle, 2007). Both compo-nents measure rehearsal, maintenance, updating, and controlled search (Unsworth &

Engle, 2007). Binding ability is then crucial in SM for the processing of items. Re-cent notions have accepted the possibility of both working memory and SM, and in doing so, have formed a blend of both theories, where a greater capacity under SM may originate from a higher WMC (Wilhelm et al., 2013). An individual’s WMC is

then believed to represent the finite ability to maintain bindings and the ability to re-solve conflicts that arise due to interference within these bindings (Oberauer, Süβ, Wilhelm, & Sander, 2007; Wilhelm et al., 2013).

Wilhelm et al. (2013) conducted a series of Confirmatory Factor Analyses (CFA) and Structural Equation Models (SEM) in order to identify the shared vari-ance between underlying executive and WMC factors that were related to Gf. For present purposes, the key aspects of this research was the inclusion of switching, in-hibiting, and updating measures derived from the executive function literature; a se-ries of working memory updating tasks such as the recall n-back task; a sese-ries of complex span tasks including the OSPAN task; and a series of binding tasks. Im-portantly, the three binding tasks involved pairing stimuli from different domains.

For the letter-colour binding task, this was a small set of letters were paired with a small set of colours. The letter-colour combinations changed from trial to trial, re-quiring the formation of new associations both within and across trials. The word-number binding task involved the pairing and repairing of a small set of words with a small set of numbers. The location-letter binding task required letters to be bound to spatial locations within a 3x3 matrix. For all three tests, the items on each trial were presented in random orders, and participants simply had to recall the pairing of a par-ticular letter with a parpar-ticular colour, a specific word with its paired number and a letter in its correct spatial location. The first outcome of the study was the confirma-tion that that the executive updating, working memory updating, and complex span latent variables where highly correlated (Wilhelm et al., 2013). This replicates the Miyake et al. (2000) finding of substantial shared variance between updating and complex span variables and provides evidence for the unity of executive functions.

This suggested that updating information is a key cognitive variable denoting indi-vidual capacities (Wilhelm et al., 2013). In the second analysis, the binding measures were entered into the model. Binding was found to be virtually identical to the execu-tive updating measure and correlated highly with the working memory updating and complex span measures (Wilhelm et al., 2013). Thereby, the final statistical model consisted of a common working memory latent variable that defined the binding and executive updating constructs.

In further analyses, they explored the executive factor of inhibition and the relationship of all previously added executive variables, and their relations to Gf.

This included inhibition tasks such as the Simon task and Erikson Flanker test. Alt-hough, the findings did not account for the executive attention explanation of WMC, where WMC is believed to inhibit interference effects, because of the lack of correla-tion between the interference errors under the incongruence condicorrela-tions of the inhibi-tion measures (Wilhelm et al., 2013). This finding therefore accounted for the diver-sity of executive functions. While there was a relationship between WMC and the tasks measuring inhibition, this relation was thought to be due to individual differ-ences in reaction times and not a reflection of cognitive control through executive at-tention (Wilhelm et al., 2013). Rather, the best explanation for the observed pattern of results was based on the notion that WMC is used for building, maintaining and updating arbitrary bindings in presence of sources of interference, and that this is be-lieved to reflect Gf (Wilhelm et al., 2013).

2.4. Executive Function, Executive Attention, and Working Memory