Reasoning with linear orders: differential parietal cortex activation in sub-clinical depression. An FMRI investigation in sub-clinical depression and controls

Abstract

The capacity to learn new information and manipulate it for efficient retrieval has long been studied through reasoning paradigms, which also has applicability to the study of social behavior. Humans can learn about the linear order within groups using reasoning, and the success of such reasoning may vary according to affective state, such as depression. We investigated the neural basis of these latter findings using functional neuroimaging. Using BDI-II criteria, 14 non-depressed (ND) and 12 mildly depressed volunteers took part in a linear-order reasoning task during functional magnetic resonance imaging. The hippocampus, parietal, and prefrontal cortices were activated during the task, in accordance with previous studies. In the learning phase and in the test phase, greater activation of the parietal cortex was found in the depressed group, which may be a compensatory mechanism in order to reach the same behavioral performance as the ND group, or evidence for a different reasoning strategy in the depressed group.

Keywords: fMRI; reasoning; sub-clinical depression.

Figure 1

Design of the study.

Figure 2

Behavioral data. (A) Mean accuracy scores (percentage of correct responses) and (B) mean response time in seconds for each test query type (one step, two step, or end point) and for each group (non-depressed or depressed).

Figure 3

Activation map for inferred queries compared to previously presented queries in ND group. Activation shown in medial frontopolar cortex (BA 10, peak −4, 64, 14, z = 3.26) and superior frontal cortex (BA8, peak −26, 32, 50, z = 3.63) in the contrast between inferred queries (two step and end point) compared to previously presented queries (one step), in the non-depressed group only. Cluster-based threshold: z > 2.3,p < 0.05.

Figure 4

Differences between D–ND groups for end-point and one-step queries. This figure shows the significantly different pattern of activation found in the parietal lobe/post-central gyrus for end-point and one-step queries between groups (D–ND): (A) for end-point queries, foci were found in superior parietal cortex (26, −46, 60, z = 3.72), supramarginal gyrus/post-central gyrus (44, −26, 40, z = 3.76); (B) shows how activation in each group during end-point queries varied according to BDI-II score; (C) for one-step queries, foci were found in the parietal lobe (post-central gyrus 60, −12, 20, z = 3.95; 58, −1, 46, z = 3.87); (D) shows how activation in each group during one-step queries varied according to BDI-II score.

Figure 5

Correlation in D group only between RT to end-point queries and activation in regions in the D–ND contrast. A scattergram plotting the reaction time during end-point queries against activation in regions in the D–ND contrast in both groups. A significant negative correlation between response time, after checking for outliers (none were found) using the Tukey criterion (Clark-Carter, , Chapter 9), and activation during end-point trials was found in the D group (r = −0.579, p = 0.048; filled diamonds), but not ND group (r = −0.009, p = 0.975; clear squares).