Generality with specificity: the dynamic field theory generalizes across tasks and time scales

Abstract

A central goal in cognitive and developmental science is to develop models of behavior that can generalize across both tasks and development while maintaining a commitment to detailed behavioral prediction. This paper presents tests of one such model, the Dynamic Field Theory (DFT). The DFT was originally proposed to capture delay-dependent biases in spatial recall and developmental changes in spatial recall performance. More recently, the theory was generalized to adults' performance in a second spatial working memory task, position discrimination. Here we use the theory to predict a specific, complex developmental pattern in position discrimination. Data with 3- to 6-year-old children and adults confirm these predictions, demonstrating that the DFT achieves generality across tasks and time scales, as well as the specificity necessary to generate novel, falsifiable predictions.

Figure 1

A simulation of the dynamic field theory performing one spatial recall trial with adult (A–C) and child (D–F) parameter settings. In each panel, location is along the x-axis, activation along the y-axis, and time along the z-axis. Solid and dashed arrows show excitatory and inhibitory connections between layers (respectively). The target was presented at 35° for 2 s, followed by a 10 s delay. The reference frame (midline) was presented continuously at 0°. See text for details.

Figure 2

A simulation of the DFT responding same in a position discrimination trial. S1 and S2 were presented at 35° for 1 s each, with a 1 s delay in between. Panels A–C show the same three fields as Figure 1 (axes and arrows are as in Figure 1), with the addition of response nodes: PF is coupled to the different (D) node, and SWM to the same (S) node; these nodes mutually inhibit one another. Panels D–G show time-slices through PF and SWM at the end of the delay (D, E) and at response (F, G). Horizontal dotted lines in these panels indicate the zero threshold; arrows specify stimulus locations, and vertical dashed lines mark the position of S1 for comparison. Note that, for simplicity, we did not include reference input in these simulations.

Figure 3

A simulation of the DFT responding different in a position discrimination trial. S1 was presented at 35°, S2 at 45°. Panels, axes, and arrows are as in Figure 2.

Figure 4

The influence of directional drift on position discrimination in the DFT using adult parameters. Panels A–F show time-slices through PF and SWM at the end of the delay (A, B) and at response when S2 was presented away from midline (C, D) or toward midline (E, F). Axes and arrows are as in Figure 2D–G. Behavioral predictions (G–I) show differences for targets near versus far from midline (x-axis), and when S2 is presented toward (solid line) versus away (dashed line) from midline. Low discrimination thresholds (along the y-axis) correspond to better performance. See text for further details.

Figure 5

The influence of directional drift on position discrimination in the DFT using child parameters. Panels, axes, and arrows are as in Figure 4.

Figure 6

Developmental predictions based on reference-related inhibition and increasing spatial precision in the DFT for discrimination performance: differences across locations and S2-direction based on only directional drift (A–D), only peak width (E–H), or both (I–L). Solid lines indicate predicted performance when S2 is presented toward midline; dashed lines indicate predicted performance when S2 is presented away from midline. Note that lower thresholds correspond to better performance. Predictions for young children (A, E, I) and adults (D, H, L) are reproduced from previous figures for comparison (see dashed boxes).

Figure 7

Testing apparatus and stimuli. Note that stimuli are presented schematically and are not drawn precisely to scale. See Apparatus section for exact stimulus details.

Figure 8

Mean discrimination thresholds (A) and mean threshold variability (B) across Targets and Ages, separately for each stimulus Direction. Solid lines indicate the S2-Toward condition; dashed lines indicate the S2-Away condition. Lower discrimination thresholds (along the y-axis in A) correspond to better performance.