Working memory and fluid intelligence: capacity, attention control, and secondary memory retrieval

Abstract

Several theories have been put forth to explain the relation between working memory (WM) and gF. Unfortunately, no single factor has been shown to fully account for the relation between these two important constructs. In the current study we tested whether multiple factors (capacity, attention control, and secondary memory) would collectively account for the relation. A large number of participants performed multiple measures of each construct and latent variable analyses were used to examine the data. The results demonstrated that capacity, attention control, and secondary memory were uniquely related to WM storage, WM processing, and gF. Importantly, the three factors completely accounted for the relation between WM (both processing and storage) and gF. Thus, although storage and processing make independent contributions to gF, both of these contributions are accounted for by variation in capacity, attention control and secondary memory. These results are consistent with the multifaceted view of WM, suggesting that individual differences in capacity, attention control, and secondary memory jointly account for individual differences in WM and its relation with gF.

Keywords: Fluid intelligence; Working memory.

Figure 1

The top row shows the schematic of the shape WM recall task. The memory array consisted of 6 shapes randomly selected from a pool of 180 shapes that vary on a circular continuum (see Zhang & Luck, 2008 for detail). When the test array was presented, participants indicated the shape at the probed location by clicking on the shape circle. They were instructed to extrapolate and click in between the shapes displayed to be as precise as possible. The middle row shows the schematic of the orientation task. The memory array presented 6 clock face stimuli. When the test array was displayed, participants indicated the orientation of the probed clock hand by clicking on the rim of the probe. They were instructed to be as precise as possible. The bottom row shows the schematic of the space task. The memory array presented 6 letters (A~F). When the test array was displayed, participants indicated the precise location of the probe letter (in this case “D”) by clicking on the grey circle.

Figure 2

Model for working memory storage (WM-S), Capacity, attention control (AC), secondary memory (SM), and fluid intelligence (gF). Paths connecting latent variables (circles) to each other represent the correlations between the constructs, the numbers from the latent variables to the manifest variables (squares) represent the loadings of each task onto the latent variable, and numbers appearing next to each manifest variable represent error variance associated with each task. All loadings and paths are significant at the p < .05 level.

Figure 3

Structural equation model for working memory storage (WM-S), Capacity, attention control (AC), secondary memory (SM), and fluid intelligence (gF). Single-headed arrows connecting latent variables (circles) to each other represent standardized path coefficients indicating the unique contribution of the latent variable. Solid lines are significant at the p < .05 level and dotted lines are not significant at the p < .05 level.

Figure 4

Model for working memory storage (WM-S), Capacity, attention control (AC), secondary memory (SM), fluid intelligence (gF), and working memory processing (WM-P). Paths connecting latent variables (circles) to each other represent the correlations between the constructs, the numbers from the latent variables to the manifest variables (squares) represent the loadings of each task onto the latent variable, and numbers appearing next to each manifest variable represent error variance associated with each task. All loadings and paths are significant at the p < .05 level.

Figure 5

Structural equation model for working memory storage (WM-S), working memory processing (WM-P), and fluid intelligence (gF). Double headed arrows connecting the latent factors represent the correlations among the factors. Single-headed arrows connecting latent variables (circles) to each other represent standardized path coefficients indicating the unique contribution of the latent variable. Solid lines are significant at the p < .05 level.

Figure 6

Structural equation model for working memory storage (WM-S), Capacity, attention control (AC), secondary memory (SM), working memory processing (WM-P), and fluid intelligence (gF). Double headed arrows connecting the latent factors represent the correlations among the factors. Single-headed arrows connecting latent variables (circles) to each other represent standardized path coefficients indicating the unique contribution of the latent variable. Solid lines are significant at the p < .05 level.

Figure 7

Venn diagrams indicating the shared and unique variance accounted for in fluid intelligence (gF), by measures of Capacity, attention control (AC), and secondary memory (SM).