Fig. 1. Subjects with Major Depressive Disorder (MDD) were oversensitive to misleading negative feedback. (A) The probabilistic switch rate (the number of reversals after misleading negative feedback/total number of probabilistic errors) was increased in the MDD group compared to healthy controls, and tended to be higher in the MDD subjects than in the subjects with Bipolar Disorder (BD), who displayed a similar degree of depression. (B) There were no differences between the three groups in the reversal switch rate (the number of reversal errors/total number of rule reversals). Error bars depict standard error of the mean.

Fig. 2. A region-of-interest analysis was used to assess differences in the hemodynamic responses, measured as the mean changes in BOLD contrast units, to the task conditions across the three diagnostic groups. The bars represent the following contrasts: Negative Feedback [events (2 + 3 + 4 + 5 + 6)− 1]; Error Switches (events 2–1); Error Non-switches (events 3–1) and the contrast Error Switches–Error Non-Switches (event 2–3). Error bars depict standard error of the mean. The regions-of-interest were defined as spheres of 10 mm radius centred around the following coordinates: (A) in dorsomedial PFC (x, y, z = 8, 32, 52), hemodynamic activity was elevated in HC and BD subjects compared to MDD subjects during error switches, and in addition the BD group had elevated activity compared to MDD and HC groups for error non-switches and to MDD for the contrast for all negative feedback; (B) in right ventrolateral PFC (38, 24, − 2), hemodynamic activity was elevated in HC and BD groups compared to the MDD group when error switches were compared to error non-switches, and when error switches were compared to correct responses; (C) in left ventrolateral PFC (MNI coordinates x, y, z = − 32, 24, − 4), hemodynamic activity was elevated in both HC and BD groups compared to MDD when error switches were compared to error non switches or correct responses.

Fig. 3. Scatter graphs depicting the relationship between the signal change in the right amygdala during receipt of negative feedback (minus the correct response baseline) and the probabilistic switch rate. The probabilistic switch rate was calculated from the total number of response reversals after misleading negative feedback, divided by the total number of trials where misleading negative feedback was presented. (A) The correlation between the amygdala response to negative feedback and the probabilistic switch rate was significant in the healthy control (HC) group (r = 0.63, p = 0.011) and (B) was apparent at trend level in the bipolar depressed group (BD) (r = 0.50, p = 0.096). (C) There was no significant association between right amygdala response and probabilistic switch rate in the major depressive disorder (MDD) group (r = − 0.16, p = 0.59), and the correlation between HC and MDD groups differed significantly (Fisher's r to z: z = 2.13 p = 0.033). (D) For the all negative feedback contrast there was less of a reduction in right amygdala activity in the MDD group compared to the HC group (p = 0.05). Error bars depict standard error of the mean.

Demographic and clinical characteristics of the healthy control and depressed subject groups and behavioral results from the probabilistic reversal learning task

Data in parentheses are S.E.M.

WASI–IQ Wechsler Abbreviated Scale of Intelligence to estimate full IQ; HSES Hollingshead and Redlich's scale of socioeconomic status; MADRS Montgomery–Asberg Depression Rating Scale; YMRS Young Mania Rating Scale; HAM-A Hamilton Anxiety Rating Scale.

Significant results for the dorsomedial and ventrolateral prefrontal cortical regions-of-interest

Regions where hemodynamic activity during the probabilistic reversal learning test was significantly higher in healthy controls than in depressed MDD subjects in the whole brain analysis

Differences between in the BOLD signal measured during each task condition relative to the baseline condition (“correct” responses) are compared across groups.

Results remain significant at p ≤ 0.05 after applying corrections for multiple comparisons using the cluster test (Poline et al., 1997); coordinates correspond to the stereotaxic array of Talairach and Tournoux (1988) and denote the distance in mm from the anterior commissure, with positive x = right of midline, positive y = anterior to the anterior commissure, and positive z = dorsal to a plane containing both the anterior and the posterior commissures. Abbreviations: C, cortex; HC, Healthy Control; MDD, Major Depressive Disorder; BD, Bipolar Disorder; G, gyrus; L, left; R, right and PFC, prefrontal cortex.

Regions where hemodynamic activity during a probabilistic reversal learning test significantly differed between depressed BD subjects and depressed MDD subjects in the whole brain analyses

Differences between in the BOLD signal measured during each task condition relative to the baseline condition (“correct” responses) are compared across groups.

Coordinates and other abbreviations interpreted as in Table 3.