Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2016 Apr 1.
Published in final edited form as: Depress Anxiety. 2015 Jan 23;32(4):296–305. doi: 10.1002/da.22338

Potentiated processing of negative feedback in depression is attenuated by anhedonia

E M Mueller 1,2, P Pechtel 2, AL Cohen 2, SR Douglas 2, DA Pizzagalli 2
PMCID: PMC4374007  NIHMSID: NIHMS642693  PMID: 25620272

Abstract

Background

Although cognitive theories of depression have postulated enhanced processing of negatively valenced information, previous EEG studies have shown both increased and reduced sensitivity for negative performance feedback in MDD. To reconcile these paradoxical findings, it has been speculated that sensitivity for negative feedback is potentiated in moderate MDD but reduced in highly anhedonic subjects. The goal of this study was to test this hypothesis by analyzing the feedback-related negativity (FRN), frontomedial theta power (FMT), and source-localized anterior midcingulate cortex (aMCC) activity after negative feedback.

Methods

Fourteen unmedicated participants with MDD and 15 control participants performed a reinforcement learning task while 128-channel EEG was recorded. FRN, FMT and LORETA source-localized aMCC activity after negative and positive feedback were compared between groups.

Results

The MDD group showed higher FRN amplitudes and aMCC activation to negative feedback than controls. Moreover, aMCC activation to negative feedback was inversely related to self-reported anhedonia. In contrast, self-reported anxiety correlated with feedback-evoked frontomedial theta (FMT) within the depression group.

Conclusions

The present findings suggest that, among depressed and anxious individuals, enhanced processing of negative feedback occurs relatively early in the information processing stream. These results extend prior work and indicate that although moderate depression is associated with elevated sensitivity for negative feedback, high levels of anhedonia may attenuate this effect.

Keywords: Feedback-related negativity, action monitoring, theta, error-related negativity, depression, LORETA

Introduction

The cognitive triad of depression includes potentiated negative processing of oneself, one's future and one's external environment [1]. Feedback on own performance is an important external source for the self-concept and hence of particular relevance for depression. Whether depressed individuals process negative feedback more intensely at relatively early stages of the information processing stream is not yet fully understood.

Early brain responses to feedback can be measured with EEG. In particular, the feedback-related negativity (FRN) [2] is a negative deflection in the event-related potential that peaks at frontomedial electrodes about 250 ms after an external feedback stimulus has been presented and is more negative for negative vs. positive feedback. Of relevance, the FRN is generated in or nearby the anterior midcingulate cortex (aMCC) [e.g. 3, 4, 5]. The FRN may reflect an affective response to negative feedback [4], is relevant for reinforcement learning [5] and has been linked to dopamine [6–9]. Because abnormal dopamine availability has been hypothesized in depression [10] and reward dysfunction in MDD [11], a link between depression and FRN is plausible from a biological point of view.

Previous studies on the relationship between FRN and depression reported increased amplitudes in response to negative feedback in moderate [12], subclinical [13] or remitted [14] depression, whereas others have linked depression to reduced FRN amplitudes in adults [15], children [16] and adolescents [17] or reported no associations between FRN to positive vs. negative feedback and depression [18]. Analyses within groups of depressed individuals indicate that high levels of symptom severity [12], particularly anhedonia [19], are associated with relatively lower FRN amplitudes to negative vs. positive feedback. In addition, the aMCC and surrounding regions are crucially implicated in the pathophysiology of MDD [20, 21] and fMRI studies have reported both aMCC hyper-[22] and hypo- [23] activity in response to negative stimuli (including negative feedback) in MDD. This pattern of mixed findings may be due to differences of paradigms and task contexts [24–26]. In addition, findings could be integrated by assuming that moderate, subclinical or remitted depression is characterized by potentiated processing of negative feedback in the aMCC, whereas very high levels of depression and anhedonia predict normal or even reduced responses to negative feedback [12,27].

In addition to FRN, the total power of frontomedial oscillations in the theta band (4–8 Hz, FMT) is enhanced after negative vs. positive feedback [28, 29]. FMT in response to negative vs. positive feedback is not correlated with the FRN across individuals [26] and has previously been related to depression in one study [30] but not in another [13]. FMT may reflect cognitive control processes with particular relevance for anxiety [31]. In line with this assumption, we recently showed increased FMT in response to (a) negative feedback in high vs. low trait anxious individuals [26] and (b) previously fear conditioned stimuli in normally anxious individuals [32]. Taken together, FMT and FRN may provide complementary electrophysiological indices of feedback processing. FMT may not be increased in depressed individuals per se but rather be associated with anxiety, which is generally elevated in depression.

The aim of the current study was to investigate FRN, FMT and source-localized aMCC activity in healthy and depressed individuals with varying degrees of anhedonic symptoms and anxiety. We hypothesized that depressed individuals would be characterized by increased FRN amplitudes and aMCC activity, although reduced sensitivity for negative feedback may be observed in the context of elevated anhedonic symptoms within the MDD group [12]. Moreover, we hypothesized that symptoms of anxiety would correlate with FMT to negative feedback. To test these hypotheses participants performed a probabilistic learning task [33] previously shown to evoke increased FRN amplitudes in individuals with subclinical depression[13].

Materials and Methods

Participants and Procedure

Twenty-nine individuals participated in this study. Subjects were only included if they were right-handed and reported no significant medical or neurological conditions, current mood disorders (other than MDD for the depressed group) or current or past psychotic symptoms, somatoform disorders, personality disorders, lifetime substance dependence, substance abuse within the past 12 months, seizures, or use of antidepressant medication in the past 2 months (6 weeks for fluoxetine). Presence (MDD group, n = 14) and absence (control group, n = 15) of depression was confirmed using the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I) [34]. Characteristics and demographic information on the sample are provided in Table 1.

Table 1.

– Demographics and Clinical Data

MDD (n = 14) Controls (n = 15) Statistical Value p-value
Demographics
 Age, mean (SD), years 28.3 (8.0) 24.5 (4.0) t(28)=1.58 .13
 Females, No 11 13 χ2=.33 .65
 White, No 10 9 χ2=.42 .70
 Years Education, mean (SD) 15.14 (2.03) 16.26 (1.83) t(28)=1.57 .13
Clinical measures
 BDI-II score, mean (SD) 25.79 (7.74) 7.67 (9.60) t(28)=5.57 <.001
 MASQ AD score, mean (SD) 83.57 (10.81) 45.87 (10.51) t(28)=9.50 <.001
 MASQ AA score, mean (SD) 24.93 (5.89) 17.93 (1.10) t(28)=4.37 <.001
Length MDE, mean (SD), month1 3.9 (2.3) -
Past episodes
 1 prior episode, No 2 0
 2 prior episodes, No 2 0
 3–5 prior episodes, No 3 0
 >6 prior episodes, No 4 0
 Chronic condition, No 5 0
 Age at first episode, mean (SD), y 18.9 (7.6) -
Distress disorder in family
 Mother, No 5 0
 Father, No 0 0
Open in a new tab
1

MDE: Major Depressive Episode. Only non-chronic MDD participants included in the mean.

Paradigm

A reinforcement-learning paradigm [35] was used, which is described in more detail elsewhere [36]. The task consists of 2–6 training blocks (60 trials per block) depending on participants' performance plus a subsequent test phase with a single block. In the training blocks three different stimulus pairs of Snodgrass images (A-B, C-D, E-F) were randomly presented on a computer screen and participants were instructed to select as quickly and accurately as possible one of the two images that they thought had the highest chance of being correct. Feedback was probabilistic with the following probabilities for positive feedback after A, B, C, D, E, or F selection: 80%, 20%, 70%, 30%, 60%, 40%. Accordingly, choosing A over B, C over D, and E over F was most often followed by positive feedback, whereas choosing B over A, D over C and F over E was most often followed by negative feedback. All participants performed at least two blocks and completed up to six blocks until a predefined learning criterion was met [36]. To ensure comparable amount of learning experience across groups, the number of blocks required to reach the learning criteria was analyzed as a behavioral measure of training performance.

Of relevance, favoring A over B can be achieved by learning that stimulus A usually leads to positive feedback (“Choose A” = learning from reward), stimulus B usually leads to negative feedback (“Avoid B” = learning from punishment), or both. To probe which type of learning had primarily occurred within individuals, the test phase consisted of not only the three previously learned or “familiar” pairs (A-B, C-D, and E-F) but also 12 “novel” combinations of all possible stimuli pairs (e.g., A–C, A–D, etc.) and no feedback was given. For the test phase, the number of accurate decisions (= choose A in A-novel and avoid B in B-novel trials) as well as reaction times for A-novel and B-novel trials was analyzed.

EEG Recording and Analysis

EEG was recorded referenced to Cz in an electrically and acoustically shielded room using a 128 channel EGI (Electrical Geodesics Inc.) hydrocell system with an NA 300 amplifier at a sampling rate of 250 Hz and a 0.1– 100 Hz bandpass filter. Impedances were below 85 kOhms.

The EEG was re-referenced to the average reference, low-pass filtered with a 50 Hz, 24 Hz/octave cutoff and manually screened for non-ocular artifacts using Brain Vision Analyzer 2 (Brain Products, Gilching, Germany). Eye-movement artifacts (e.g., blinks) were removed using independent component analysis (ICA), and EEG channels with a high number of channel-specific artifacts were removed and topographically interpolated using spline interpolation. The EEG of the training phase was segmented from −500 to 1000 ms, baseline-corrected (−200 to 0 ms), averaged across all trials in which positive or negative feedback was given and corrected for a fixed 36 ms delay relative to real time that was caused by the anti-aliasing filter of the EEG recording system. In order to detect the time points with maximal differentiation between positive and negative feedback at channel FCz (E6), a grand average difference wave FRN (negative minus positive feedback) across groups was computed (Figure 2). Because the difference wave peaked at 224 ms, the FRN was measured as the mean amplitude at 224 ± 50 ms at a cluster of electrodes around channel FCz (E5, E6, E7, E12, E13, E106, E112). To test for the specificity of our hypothesized results with regard to the FRN we also measured the P300 as the difference wave, which revealed a maximum differentiation of positive vs. negative feedback at 364 ms. Accordingly, the P300 was measured at 364 ± 801 ms at a cluster of electrodes around channel Cz (E7, E10, E31, E55, E80), where the topography of the difference wave was maximal (Figure 2A and 2B). To measure FMT, a Fast Fourier Transform (FFT) was chosen because FFT provides a robust aggregate measure of spectral power within a pre-defined time period that is sensitive to feedback valence [26]. Segments from 0 to 1 s relative to feedback onset were 50% hamming windowed (thereby attenuating the impact of very early and late EEG during the segment) and subjected to FFT analysis [26]. Prior to statistical analyses, power in the theta range (4–8 Hz) was log-transformed. FMT analyses were performed for the same channels as the FRN analyses.

Figure 2.

Figure 2

Open in a new tab

(A) Barplot indicating mean (SEM) reaction times for A-novel (white) and B-novel (gray) trials. A reaction time slowing for B-novel vs. A-novel trials can be observed across groups but is more pronounced in the MDD group. (B) Scatterplot showing the relationship between reaction time slowing (y-axis) and MASQ Anxious Arousal (x-axis) within the MDD group.

LORETA

To localize brain sources of putative group differences in the scalp data, we used the LORETA algorithm [37]. The LORETA transformation matrix was computed based on the positions of the hydrocell geodesic sensor net EEG-sensors in Talairach space without over-smoothing. Current source density was then estimated based on the average ERP amplitude from 174 to 274 ms relative to feedback onset (FRN) at all 129 sensors. For each subject and condition, LORETA activity was then normalized to unity and log-transformed.

Symptoms of anhedonic depression and anxiety

To separately assess symptoms of depression and anxiety within individuals with a diagnosis of MDD, all participants completed the Mood and Anxiety Symptom Questionnaire, a reliable instrument that allows dissociating symptoms of depression and anxiety [38]. Anhedonia was measured with the Anhedonic Depression (AD) subscale, whereas anxiety was assessed with the anxious apprehension (AA) subscale. In support of the two scales probing separate constructs, correlations of the two scales are typically in the medium range (i.e., between .2 and .5 [38]) and showed a correlation of .42 in the present MDD sample. Participants also completed the Beck Depression Inventory II [39], among other questionnaires.

Statistical Analyses

Hypotheses on group differences were tested using repeated measures Analyses of Variance (ANOVAs) with Greenhouse-Geisser corrections were applicable. For FRN and FMT, the between subjects factor was Group (Control vs. MDD) and the within-subjects factors were Electrode and FeedbackValence (Training Phase: Positive vs. Negative Feedback) and the critical p-value was set to .05. For source estimated activity evoked by positive and negative feedback, voxel-wise independent samples t-tests were performed with a statistical threshold of .005 and a minimal cluster threshold of 5 contiguous significant voxels. Within groups, contrasts capturing effects of interest were correlated with different questionnaire measures. To test for differences between dependent correlation coefficients, we used the approach suggested by Meng et al. [40].

Results

Behavioral results

Groups did not differ in the number of blocks needed to reach the learning criterion in the training phase (control group: 2.9 ± 1.2 Blocks, MDD: 3.6 ± 1.7 Blocks; t(27)=1.15, P > .2)2 indicating that for both groups comparable numbers of trials were available for EEG analysis (Figure 1).

Figure 1.

Figure 1

Open in a new tab

Accuracy for A–B (blue), C–D (red) and E–F (green) trials for each training block and the test phase (x-axis) for the control group (left) and MDD group (right). The size of the block indicates the number of subjects in the corresponding block due to not reaching the learning criterion in the previous block. Therefore, decreasing square sizes as a function of training block indicate that more participants reached the learning criterion. Performance during the test phase for A–B (blue X), A-novel (blue triangle) and B-novel (blue circle) trials is also indicated. The dashed black line indicates the chance level. As evident from the figure, both the control and MDD group showed above-chance selection of A in A-novel trials and (correct) avoidance of B in B-novel trials during the test phase, indicating successful transfer from training to test phase.

For the test phase the Group x Condition (“A-novel” vs. “B-novel”) ANOVA on accuracy revealed no significant effects (all Ps > .09). The Group x Condition ANOVA on reaction time revealed delayed responses during B-novel vs. A-novel trials (main effect Condition: F(1,27) = 16.01, P < .001). The reaction time difference for B-novel vs. A-novel trials was larger in MDD vs. control participants (Group x Condition interaction: F(1,27) = 5.80, P < .025; Figure 2A) and correlated with the MASQ AA (r = .56, P < .04, Figure 2B) but not AD scale (r = .02, n.s.; comparison of correlations: Z = 1.78, P < .04, one-tailed) within the MDD group, suggesting that more anxious depressed individuals responded slower when previously punished stimuli presented.

Electrophysiological results

FRN (224 ± 50 ms)

As expected the Group x Feedback x Electrode ANOVA on the mean FRN amplitude revealed a significant main effect of Feedback (F(1,27)=22.84, P < . 0001) indicating more negative amplitudes for negative vs. positive feedback. Importantly, there was a significant Group x Feedback interaction (F(1,27)=5.99, P < .025), which was due to increased differentiation between positive and negative feedback in the MDD (F(1,14)= 51.35, P = .0001) vs. control (F(1,14)= 1.91, P = .19) group (Figures 3C and 3D). Finally, there was a Group x Feedback x Electrode interaction (F(1,27)=2.43, P > . 05) indicating that the Group x Feedback interaction was most pronounced at centromedial electrode E13 (F(1,27)= 10.33, P < .003).

Figure 3.

Figure 3

Open in a new tab

(A) Grand average difference wave for positive minus negative feedback-evoked event-related potentials at channel FCz and a cluster of surrounding electrodes. This difference wave across all participants was used to identify the time window for the FRN (224 ms ± 50 ms) and for the P300 (364 ± 80 ms) based on maximum differentiation between positive and negative feedback independent of group. (B) Topographies of the positive minus negative feedback contrast during the FRN and P300 time window. (C and D) Event-related potentials at channel FCz and a cluster of surrounding electrodes for positive (gray line) and negative (black line) feedback, separately for the control (C) and MDD (D) group. The measurement windows for FRN (light blue) and P300 (light red) are also indicated.

The positive vs. negative feedback difference FRN averaged across electrodes was not correlated with the MASQ AD or MASQ AA scales within the depression group (both Ps > .3).

LORETA (224 ± 50 ms)

As shown in Figure 4A, whole-brain group comparisons revealed that MDD participants displayed more rostral aMCC activation to negative feedback than control participants (P < .005) while there were no significant group differences in the aMCC or surrounding regions after positive feedback (see also Figure 4C). Of particular relevance, after negative feedback, the LORETA-estimated activity in the aMCC-region with the maximum group difference (Talairach coordinates: X: −10, Y: 31, Z: 29) was negatively correlated with the MASQ AD scale (r = −0.64, P < .015; Figure 3D). The AA scale (r = −0.44, p > .1) and the BDI (r = −0.25, P > .3) were not significantly correlated with aMCC activity to negative feedback. Directly comparing these three correlation coefficients yielded a significant difference for AD-aMCC vs. BDI-aMCC correlations (Z = 2.34, P < .01) [40] but not AA-aMCC vs. AD-aMCC correlations (P >.2) suggesting that anhedonia symptoms (rather than general depression severity) related to blunted responses to negative feedback in depression. Within the control group, the MASQ AD (r = −.25) and AA (r = .11) scale and the BDI (r = −.12) were not significantly correlated with aMCC activity to negative feedback (Ps > .3).

Figure 4.

Figure 4

Open in a new tab

(A) Whole brain analyses of significant group differences (P < .005) 224 ± 50 ms after positive (light colors) or negative (dark colors) feedback presentation. Red indicates greater current source density for control vs. MDD participants, green indicates greater current source density for MDD vs. control participants. (B) Barplot indicating mean (±SEM) log-transformed normalized activity in the anterior midcingulate cortex region shown in (A), separately for group (control group: red; MDD group: green) and feedback (positive: light colors; negative: dark colors). (C) Scatterplot of Anhedonia symptoms (x-axis) and log-transformed normalized activity in the anterior midcingulate cortex region shown in B (r = −0.64, P < .015). The mean (red square) and SD (error bars) of the control group for both variables are also shown.

P300 (364 ± 80 ms)

The Group x Feedback x Electrode ANOVA on the mean P300 amplitude revealed increased P300 amplitudes for negative vs. positive feedback (main effect of Feedback, F(1,27) = 20.80, P < .001) but no significant main effects or interactions involving Group (Ps > .4), thereby supporting the specificity of the FRN-related findings reported above (see also Figure 3C).

Feedback-evoked theta

The Group x Feedback x Electrode ANOVA on theta power revealed increased theta power to negative vs. positive feedback, irrespective of group (main effect Feedback: (F(1,29) = 17.93, P < .0001; Feedback x Group interaction: P > .5). The difference of negative vs. positive feedback-evoked theta power and the difference-FRN were not correlated (r = .11, P >.5) and theta was also not correlated with the difference P300 (r = .11, P > .5), indicating that all three measures capture different aspects of individual differences in feedback processing [26].

Importantly, among the MDD participants, FMT for negative minus positive feedback showed a positive correlation with MASQ AA (r = −0.61, P < .025; Figure 5B) but not AD or BDI scores (P > .5), indicating a specific association of FMT and anxiety symptoms within the MDD group (comparison of correlation coefficients: AA-FMT vs. AD-FMT: Z = 1.57, P = .058, and AA-FMT vs. BDI-FMT: Z = 2.2, P < .02 [40].

Figure 5.

Figure 5

Open in a new tab

(A) Topography map of differences in theta power for negative vs. positive feedback. (B) Scatterplot showing the differences in theta power for negative vs. positive feedback (y-axis) as a function of anxiety (x-axis) within the MDD group.

Discussion

The aim of the current study was to investigate the processing of negative feedback in depression. Based on prior work it was hypothesized that depression would be characterized by increased frontomedial brain responses to negative feedback. Within depressed individuals however, previous findings lead us to assume that high levels of anhedonia and/or general depression severity would be associated with reduced rather than enhanced neural sensitivity for negative feedback. Finally, we expected that theta evoked by negative feedback would relate to high levels of anxiety rather than anhedonia or general depression symptoms. Using FRN, FMT and source localization of feedback-evoked brain activity our hypotheses were largely confirmed. Relative to controls, unmedicated depressed participants showed increased FRN amplitudes and increased aMCC activity to negative (versus positive) feedback. Moreover, within the MDD group, high anhedonia predicted reduced aMCC activity to negative feedback. Anxiety, in contrast, predicted not only enhanced theta responses to negative feedback but also prolonged reaction times to stimuli previously associated with negative feedback (i.e., B-stimuli). Together, these findings replicate, integrate and refine a number of earlier findings on feedback processing in depression.

In general, individuals with MDD showed significantly larger FRN to negative vs. positive feedback. Source analyses further showed that aMCC activity in the same time window was stronger in depressed versus control participants after negative (but not positive) feedback. These findings are consistent with prior studies on the FRN and similar components (i.e., error-related negativity [41]) showing increased amplitudes in depressed vs. control participants [12, 42, 43] and with studies that localized these components in or in proximity to the aMCC [4, 44, 45]. Consistent with accounts that emphasize potentiated processing of negatively valenced information in depression, these findings suggest that depressed individuals show an increased sensitivity for negative feedback at relatively early stages of information processing (i.e. ≤ 250 ms after stimulus presentation).

Of further relevance, aMCC activity to negative feedback correlated negatively with anhedonia severity. Thus, increasing levels of anhedonia were associated with relative blunted neural response to negative feedback. This result converges with prior work showing that severe depression [12, 46, 47] and anhedonia [19] are characterized by normative or even reduced processing of negative feedback and mistakes. Collectively, these findings suggest an inverted U-shaped association between depression severity and ERN/FRN amplitudes similar to what has been reported for the relationship between FRN [8] or ERN [48] amplitudes and the neuromodulator dopamine. Because dopamine has been linked to depression [10] and particularly reward processing dysfunction [11] and anhedonia [11, 49], a shared underlying mechanism is plausible. Of particular practical relevance, such curvilinear relationships would indicate that enhanced sensitivity for negative feedback is only a sensitive marker for moderate MDD without high levels of anhedonia.

Like the FRN, theta power at frontal midline electrodes was sensitive to the feedback valence. Although the FRN can be described as a phase-locked frontomedial response in the theta range, FMT as measured in the present study was based on a longer time period (i.e., 1 s) and also captures non-phase locked theta power. Accordingly, the effect of feedback on FMT was uncorrelated with the effect of feedback on the FRN (see also [26]). Because this result suggests that FRN and FMT do not capture identical brain responses, future studies should attempt to parse the underlying functions of FRN and FMT (e.g. affective responses and/or adaptations in response to negative feedback). This goal is of particular relevance given that, in contrast to FRN, FMT was not significantly enhanced in depressed individuals in the present study and has even been found to be reduced in depression with moderate levels of anhedonia [30]. While there was no association between depression and FMT, self-reported anxiety symptoms within depressed individuals were positively correlated with FMT responses to negative vs. positive feedback. This latter finding converges with earlier work showing that FMT responses to negative vs. positive feedback correlate with trait anxiety in healthy participants [26].

Longer reaction times were observed whenever B-stimuli (previously associated with negative feedback) rather than A-stimuli (previously associated with positive feedback) were presented. This effect was observed across all individuals but was significantly stronger in the MDD vs. control group and was particularly strong in MDD subjects with high anxiety. Under the assumption that these prolonged reactions to B-stimuli are due to inhibitory processes, this result may reflect anxiety-related behavioral inhibition to previously punished stimuli [50]. Taken together, anxious depressed individuals thus show increased FMT responses to negative feedback and also increased behavioral adaptation to negative feedback associated stimuli. These patterns are consistent with the hypothesis by Cavanagh and Shackman [31], which states that frontal midline theta reflects cognitive control processes with particular relevance for anxiety. Whether anxiety disorders are characterized by elevated FMT is a relevant question for future research.

An important limitation of the present study is its relatively small sample size, particularly for correlation analyses within groups. As a result it is difficult to know whether non-significant results reflect an absence of effects in the population or should instead be ascribed to low power. Moreover, the likelihood that spurious correlations reach significance is enhanced in small samples. For example, we cannot rule out that FRN in response to negative feedback may also have correlated with anxiety symptoms [51–53] or that FRN in response to positive feedback may also have correlated with anhedonia in a larger sample [19]. Future studies with larger samples are warranted to replicate and extend the present findings in order to draw firm conclusions.

Conclusion

In conclusion, the present study replicated and integrated prior studies on feedback processing in MDD. We found that increased FRN and aMCC activity to negative feedback characterize depression, whereas high levels of anhedonia seem to counteract this depression-related enhancement of negative feedback processing. Together with previous work, these findings suggest early and automatic information processing biases in moderate depression with altered processing of negative feedback. However, future studies in larger samples and implementing multiple tasks are needed to clarify how task characteristics and context may further affect feedback processing in depression.

Table 2.

Mean electrophysiological responses (+SEM) to feedback and correlations with symptom scales

MDD (n = 14) t-Value (df = 28) p-value rAD (MDD/CT) rAA (MDD/CT) rBDI (MDD/CT)
FRN
 Positive Feedback 3.27(.90) 1.49 .15 −.27/.07 −.23/.05 −.05/−.29
 Negative Feedback 1.5(.78) .59 .56 −.25/.14 −.20/.27 −.05/−.24
 Difference (pos minus neg) 1.75(.24) 2.45 .02 −.21/−.11 −.23/−.38 −.02/−.16
aMCC (TAL: -10, 31, 29)
 Positive Feedback −1.06(.11)1 1.25 .22 .15/−.25 .24/.07 .29/.32
 Negative Feedback −.97(.08) 3.39 .002 −.64*/−.25 −.44/.11 −.25/−.13
 Difference (pos minus neg) .09(.14) 1.59 .12 −.47§/−.07 −.42/.06 −.35/−.29
FMT
 Positive Feedback −.26(.12) .12 n.s. .06/.04 −.07/.11 .20/.07
 Negative Feedback −.12(.11) .05 n.s. .12/−.02 .14/.09 .20/.10
 Difference (pos minus neg) −.13(.04) .17 n.s. −.16/.16 −.61*/.01 .04/−.09
Open in a new tab

Notes:

1

FMT-values are negative due to log-transformation; rAD/AA/BDI: correlation of variable with the Anhedonic Depression scale of the MASQ, Anxious Arousal scale of the MASQ or with Beck Depression Inventory score (separate correlations for MDD and Controls are provided).

*

p < .05;

§

p < . 10

Acknowledgments

The authors are grateful to Laura Murray and Nancy Brooks Hall for their skilled assistance with this study. This study was partially supported by National Institute of Mental Health (NIMH) grant R01 MH068376 awarded to DAP.

Footnotes

Financial Disclosures Over the past three years, Dr. Pizzagalli has received honoraria/consulting fees from Advanced Neuro Technology North America, AstraZeneca, Ono Pharma USA, Pfizer, Servier, and Shire for activities unrelated to this project. All other authors report no biomedical financial interests to disclose.

1

This 80 ms range was chosen in order to also capture the raw-wave P300 peak, which occurred earlier than the difference-wave P300 peak. However, results were comparable, when the 50 ms range was used instead (i.e., 364 ± 50 ms).

2

However, as shown in Figure 1, a subgroup of n = 3 MDD participants failed to reach the learning criterion for EF-trials even after the 6th block (as opposed to 1 participant in the control group). To keep the depression group as large and representative as possible, we decided not to exclude this subgroup of participants. Interestingly, a closer analysis of this subgroup revealed that, relative to the remaining MDD group, these three participants were characterized by elevated BDI-Melancholia subscores (see [54] for BDI items contributing to this subscore), longer reaction times (291 ms longer) and lower income. This suggests that a subgroup of depressed participants with a particularly strong impairment was unable to learn from subtle rewards in the EF condition.

References

  • 1.Beck AT. Cognitive therapy and the emotional disorders. Penguin; New York: 1979. [Google Scholar]
  • 2.Miltner WHR, Braun CH, Coles MGH. Event-related brain potentials following incorrect feedback in a time-estimation task: Evidence for a “generic” neural system for error detection. Journal of Cognitive Neuroscience. 1997;9:788–98. doi: 10.1162/jocn.1997.9.6.788. [DOI] [PubMed] [Google Scholar]
  • 3.Vogt BA. Pain and emotion interactions in subregions of the cingulate gyrus. Nat Rev Neurosci. 2005;6:533–44. doi: 10.1038/nrn1704. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Gehring WJ, Willoughby AR. The medial frontal cortex and the rapid processing of monetary gains and losses. Science. 2002;295:2279–82. doi: 10.1126/science.1066893. [DOI] [PubMed] [Google Scholar]
  • 5.Walsh MM, Anderson JR. Learning from experience: event-related potential correlates of reward processing, neural adaptation, and behavioral choice. Neurosci Biobehav Rev. 2012;36:1870–84. doi: 10.1016/j.neubiorev.2012.05.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Holroyd CB, Coles MG. The neural basis of human error processing: reinforcement learning, dopamine, and the error-related negativity. Psychol Rev. 2002;109:679–709. doi: 10.1037/0033-295X.109.4.679. [DOI] [PubMed] [Google Scholar]
  • 7.Ullsperger M. Genetic association studies of performance monitoring and learning from feedback: the role of dopamine and serotonin. Neurosci Biobehav Rev. 2010;34:649–59. doi: 10.1016/j.neubiorev.2009.06.009. [DOI] [PubMed] [Google Scholar]
  • 8.Mueller EM, Burgdorf C, Chavanon ML, Schweiger D, Hennig J, Wacker J, et al. The COMT Val158Met polymorphism regulates the effect of a dopamine antagonist on the feedback-related negativity. Psychophysiology. 2014;51:805–9. doi: 10.1111/psyp.12226. [DOI] [PubMed] [Google Scholar]
  • 9.Santesso DL, Evins AE, Frank MJ, Schetter EC, Bogdan R, Pizzagalli DA. Single dose of a dopamine agonist impairs reinforcement learning in humans: Evidence from event-related potentials and computational modeling of striatal-cortical function. Hum Brain Mapp. 2008;30:1963–76. doi: 10.1002/hbm.20642. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Dunlop BW, Nemeroff CB. The role of dopamine in the pathophysiology of depression. Arch Gen Psychiatry. 2007;64:327–37. doi: 10.1001/archpsyc.64.3.327. [DOI] [PubMed] [Google Scholar]
  • 11.Pizzagalli DA. Depression, stress, and anhedonia: toward a synthesis and integrated model. Annual review of clinical psychology. 2014;10:393–423. doi: 10.1146/annurev-clinpsy-050212-185606. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Tucker DM, Luu P, Frishkoff G, Quiring J, Poulsen C. Frontolimbic response to negative feedback in clinical depression. J Abnorm Psychol. 2003;112:667–78. doi: 10.1037/0021-843X.112.4.667. [DOI] [PubMed] [Google Scholar]
  • 13.Cavanagh JF, Bismark AJ, Frank MJ, Allen JJ. Larger Error Signals in Major Depression are Associated with Better Avoidance Learning. Frontiers in psychology. 2011;2:331. doi: 10.3389/fpsyg.2011.00331. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Santesso DL, Steele KT, Bogdan R, Holmes AJ, Deveney CM, Meites TM, et al. Enhanced negative feedback responses in remitted depression. Neuroreport. 2008;19:1045–8. doi: 10.1097/WNR.0b013e3283036e73. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Foti D, Hajcak G. Depression and reduced sensitivity to non-rewards versus rewards: Evidence from event-related potentials. Biol Psychol. 2009;81:1–8. doi: 10.1016/j.biopsycho.2008.12.004. [DOI] [PubMed] [Google Scholar]
  • 16.Bress JN, Smith E, Foti D, Klein DN, Hajcak G. Neural response to reward and depressive symptoms in late childhood to early adolescence. Biol Psychol. 2012;89:156–62. doi: 10.1016/j.biopsycho.2011.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Foti D, Kotov R, Klein DN, Hajcak G. Abnormal neural sensitivity to monetary gains versus losses among adolescents at risk for depression. Journal of abnormal child psychology. 2011;39:913–24. doi: 10.1007/s10802-011-9503-9. [DOI] [PubMed] [Google Scholar]
  • 18.Mies GW, van der Veen FM, Tulen JH, Birkenhager TK, Hengeveld MW, van der Molen MW. Drug-free patients with major depression show an increased electrophysiological response to valid and invalid feedback. Psychol Med. 2011;41:2515–25. doi: 10.1017/S0033291711000778. [DOI] [PubMed] [Google Scholar]
  • 19.Liu WH, Wang LZ, Shang HR, Shen Y, Li Z, Cheung EF, et al. The influence of anhedonia on feedback negativity in major depressive disorder. Neuropsychologia. 2014;53:213–20. doi: 10.1016/j.neuropsychologia.2013.11.023. [DOI] [PubMed] [Google Scholar]
  • 20.Anand A, Li Y, Wang Y, Wu J, Gao S, Bukhari L, et al. Activity and connectivity of brain mood regulating circuit in depression: a functional magnetic resonance study. Biol Psychiatry. 2005;57:1079–88. doi: 10.1016/j.biopsych.2005.02.021. [DOI] [PubMed] [Google Scholar]
  • 21.Drevets WC. Neuroimaging studies of mood disorders. Biol Psychiatry. 2000;48:813–29. doi: 10.1016/s0006-3223(00)01020-9. [DOI] [PubMed] [Google Scholar]
  • 22.Groenewold NA, Opmeer EM, de Jonge P, Aleman A, Costafreda SG. Emotional valence modulates brain functional abnormalities in depression: evidence from a meta-analysis of fMRI studies. Neurosci Biobehav Rev. 2013;37:152–63. doi: 10.1016/j.neubiorev.2012.11.015. [DOI] [PubMed] [Google Scholar]
  • 23.Steele JD, Kumar P, Ebmeier KP. Blunted response to feedback information in depressive illness. Brain. 2007;130:2367–74. doi: 10.1093/brain/awm150. [DOI] [PubMed] [Google Scholar]
  • 24.Olvet DM, Hajcak G. The effect of trial-to-trial feedback on the error-related negativity and its relationship with anxiety. Cogn Affect Behav Neurosci. 2009;9:427–33. doi: 10.3758/CABN.9.4.427. [DOI] [PubMed] [Google Scholar]
  • 25.Tops M, Boksem MA. Absorbed in the task: Personality measures predict engagement during task performance as tracked by error negativity and asymmetrical frontal activity. Cogn Affect Behav Neurosci. 2010;10:441–53. doi: 10.3758/CABN.10.4.441. [DOI] [PubMed] [Google Scholar]
  • 26.Mueller EM, Burgdorf C, Chavanon ML, Schweiger D, Wacker J, Stemmler G. Dopamine modulates frontomedial failure processing of agentic introverts versus extraverts in incentive contexts. Cogn Affect Behav Neurosci. 2014;14:756–68. doi: 10.3758/s13415-013-0228-9. [DOI] [PubMed] [Google Scholar]
  • 27.Endrass T, Ullsperger M. Specificity of performance monitoring changes in obsessive-compulsive disorder. Neurosci Biobehav Rev. 2014 doi: 10.1016/j.neubiorev.2014.03.024. [DOI] [PubMed] [Google Scholar]
  • 28.Cavanagh JF, Zambrano-Vazquez L, Allen JJ. Theta lingua franca: a common mid-frontal substrate for action monitoring processes. Psychophysiology. 2012;49:220–38. doi: 10.1111/j.1469-8986.2011.01293.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Cohen MX, Elger CE, Ranganath C. Reward expectation modulates feedback-related negativity and EEG spectra. Neuroimage. 2007;35:968–78. doi: 10.1016/j.neuroimage.2006.11.056. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Mueller EM, Panitz C, Pizzagalli DA, Hermann C, Wacker J. Personality and Individual Differences in press. Midline theta dissociates agentic extraversion and anhedonic depression. (Doi:10.1016/j.paid.2014.10.043) [Google Scholar]
  • 31.Cavanagh JF, Shackman AJ. Journal of physiology. Paris: 2014. Frontal midline theta reflects anxiety and cognitive control: Meta-analytic evidence. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Mueller EM, Panitz C, Hermann C, Pizzagalli DA. Prefrontal Oscillations during Recall of Conditioned and Extinguished Fear in Humans. J Neurosci. 2014;34:7059–66. doi: 10.1523/JNEUROSCI.3427-13.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Frank MJ, Woroch BS, Curran T. Error-related negativity predicts reinforcement learning and conflict biases. Neuron. 2005;47:495–501. doi: 10.1016/j.neuron.2005.06.020. [DOI] [PubMed] [Google Scholar]
  • 34.First M, Spitzer R, Gibbon M, Williams J. Structured Clinical Interview for DSM-IV-TR Axis I Disorders research version non-patient edition (SCID-I/NP) Biometrics Research New York State Psychiatric Institute; New York: 2002. [Google Scholar]
  • 35.Frank MJ, Seeberger LC, O'Reilly RC. By carrot or by stick: cognitive reinforcement learning in parkinsonism. Science. 2004;306:1940–3. doi: 10.1126/science.1102941. [DOI] [PubMed] [Google Scholar]
  • 36.Pechtel P, Pizzagalli DA. Disrupted reinforcement learning and maladaptive behavior in women with a history of childhood sexual abuse: a high-density event-related potential study. JAMA psychiatry. 2013;70:499–507. doi: 10.1001/jamapsychiatry.2013.728. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Pascual-Marqui RD, Lehmann D, Koenig T, Kochi K, Merlo MC, Hell D, et al. Low resolution brain electromagnetic tomography (LORETA) functional imaging in acute, neuroleptic-naive, first-episode, productive schizophrenia. Psychiatry Res. 1999;90:169–79. doi: 10.1016/s0925-4927(99)00013-x. [DOI] [PubMed] [Google Scholar]
  • 38.Watson D, Weber K, Assenheimer JS, Clark LA, Strauss ME, McCormick RA. Testing a tripartite model: I. Evaluating the convergent and discriminant validity of anxiety and depression symptom scales. J Abnorm Psychol. 1995;104:3–14. doi: 10.1037//0021-843x.104.1.3. [DOI] [PubMed] [Google Scholar]
  • 39.Beck AT, Steer RA, Ball R, Ranieri W. Comparison of Beck Depression Inventories -IA and -II in psychiatric outpatients. J Pers Assess. 1996;67:588–97. doi: 10.1207/s15327752jpa6703_13. [DOI] [PubMed] [Google Scholar]
  • 40.Meng X-l, Rosenthal R, Rubin DB. Comparing correlated correlation coefficients. Psychological Bulletin. 1992;111:172–5. [Google Scholar]
  • 41.Gehring WJ, Coles MG, Meyer DE, Donchin E. A brain potential manifestation of error-related processing. Electroencephalogr Clin Neurophysiol Suppl. 1995;44:261–72. [PubMed] [Google Scholar]
  • 42.Chiu PH, Deldin PJ. Neural evidence for enhanced error detection in major depressive disorder. Am J Psychiatry. 2007;164:608–16. doi: 10.1176/ajp.2007.164.4.608. [DOI] [PubMed] [Google Scholar]
  • 43.Holmes AJ, Pizzagalli DA. Spatiotemporal dynamics of error processing dysfunctions in major depressive disorder. Arch Gen Psychiatry. 2008;65:179–88. doi: 10.1001/archgenpsychiatry.2007.19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Debener S, Ullsperger M, Siegel M, Fiehler K, von Cramon DY, Engel AK. Trial-by-trial coupling of concurrent electroencephalogram and functional magnetic resonance imaging identifies the dynamics of performance monitoring. J Neurosci. 2005;25:11730–7. doi: 10.1523/JNEUROSCI.3286-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Herrmann MJ, Rommler J, Ehlis AC, Heidrich A, Fallgatter AJ. Source localization (LORETA) of the error-related-negativity (ERN/Ne) and positivity (Pe) Brain Res Cogn Brain Res. 2004;20:294–9. doi: 10.1016/j.cogbrainres.2004.02.013. [DOI] [PubMed] [Google Scholar]
  • 46.Georgiadi E, Liotti M, Nixon NL, Liddle PF. Electrophysiological evidence for abnormal error monitoring in recurrent major depressive disorder. Psychophysiology. 2011;48:1192–202. doi: 10.1111/j.1469-8986.2011.01198.x. [DOI] [PubMed] [Google Scholar]
  • 47.Schrijvers D, de Bruijn ER, Maas Y, De Grave C, Sabbe BG, Hulstijn W. Action monitoring in major depressive disorder with psychomotor retardation. Cortex; a journal devoted to the study of the nervous system and behavior. 2008;44:569–79. doi: 10.1016/j.cortex.2007.08.014. [DOI] [PubMed] [Google Scholar]
  • 48.Mueller EM, Makeig S, Stemmler G, Hennig J, Wacker J. Dopamine Effects on Human Error Processing Depend on Catechol-O-Methyltransferase VAL158MET Genotype. J Neurosci. 2011;31:15818–25. doi: 10.1523/JNEUROSCI.2103-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Wise RA. Dopamine and reward: the anhedonia hypothesis 30 years on. Neurotoxicity research. 2008;14:169–83. doi: 10.1007/BF03033808. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Gray JA, McNaughton N. The Neuropsychology of Anxiety: An Enquiry into the Functions of the Septohippocampal System (2nd ed.) 2 ed Oxford University Press; Oxford (UK): 2000. [Google Scholar]
  • 51.Mueller EM, Stemmler G, Hennig J, Wacker J. 5-HTTLPR and anxiety modulate brain-heart covariation. Psychophysiology. 2013;50:441–53. doi: 10.1111/psyp.12016. [DOI] [PubMed] [Google Scholar]
  • 52.Santesso DL, Bogdan R, Birk JL, Goetz EL, Holmes AJ, Pizzagalli DA. Neural responses to negative feedback are related to negative emotionality in healthy adults. Social cognitive and affective neuroscience. 2012;7:794–803. doi: 10.1093/scan/nsr054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Sato A, Yasuda A, Ohira H, Miyawaki K, Nishikawa M, Kumano H, et al. Effects of value and reward magnitude on feedback negativity and P300. Neuroreport. 2005;16:407–11. doi: 10.1097/00001756-200503150-00020. [DOI] [PubMed] [Google Scholar]
  • 54.Pizzagalli DA, Jahn AL, O'Shea JP. Toward an objective characterization of an anhedonic phenotype: a signal-detection approach. Biol Psychiatry. 2005;57:319–27. doi: 10.1016/j.biopsych.2004.11.026. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES