Recent preliminary evidence indicates that depression is associated with impaired intuitive information processing. The current study aimed at replicating these findings and to move one step further by exploring whether factors known as triggering intuition (positivity, processing fluency) also affect intuition in patients with depression.
We pre-registered and tested five hypotheses using data from 35 patients with depression and 35 healthy controls who performed three versions of the Judgment of Semantic Coherence Task (JSCT, Bowers et al., 1990). This task operationalizes intuition as the inexplicable and sudden detection of semantic coherence.
Results revealed that depressed patients and healthy controls did not differ in their general intuitive performance (Hypothesis 1). We further found that fluency did not significantly affect depressed patients’ coherence judgments (H2a) and that the assumed effect of fluency on coherence judgments was not moderated by depression (H2b). Finally, we found that triads positive in valence were more likely to be judged as coherent as compared to negative word triads in the depressed sample (H3a), but this influence of positive (vs. negative) valence on coherence judgments did not significantly differ between the two groups (H3b).
Overall the current study did not replicate findings from previous research regarding intuitive semantic coherence detection deficits in depression. However, our findings suggest that enhancing positivity in depressed patients may facilitate their ability to see meaning in their environment and to take intuitive decision.
The pre-registered replication study did not find intuition deficits in patients with depression. Processing fluency did not affect coherence judgments in depressed patients or healthy controls. Depressed patients and healthy controls use positive valence as cue for intuitive coherence judgments. Future studies should test whether enhancing positivity in depressed patients boosts their ability to find meaning (e.g., meaning in life).
People continuously make decisions and judgments without long consideration by relying on their gut feelings. Following one’s intuition does not only
Intuition relies on processes that are based on experience, run quickly and unconsciously, and allow many relevant aspects to be effortlessly integrated into a coherent whole (
It seems reasonable to assume that people are not always able to intuitively detect meaning and coherence. One important influencing factor seems to be the affective state of a person. Positive mood broadens the scope of attention, facilitates associative processing (
Depression is characterized by negative mood and a brooding, rigid, style of thinking. This is opposed to an intuitively integrating and holistic style of processing (see
Apart from the influence of people’s mood states (
In addition, there is also evidence showing that manipulating both fluency and positive affect influences whether people feel coherence. Manipulating positivity on a subtle level (e.g., by subliminal affective facial priming or by manipulating the affective valence of word triads or solution words;
The aim of the current study was to replicate and extend preliminary evidence on intuition deficits in depression. We tested a sample of depressed inpatients and compared their performance in the Judgment of Semantic Coherence Task (JSCT;
The following main hypotheses were pre-registered and investigated (see the
The second hypothesis (H2a) assumes that processing fluency triggers semantic coherence judgments in patients with depression. Building up on basic research, we expected that in depressed patients word triads that are presented in a high figure-ground contrast – and which are therefore presumed to be processed more fluently – are more likely to be judged to be coherent than triads presented in a low contrast. Given that research using self-reports (
The third hypothesis (H3a) was that the positive valence of stimuli influences semantic coherence judgments in patients with depression. Building up on basic research (
Forty inpatients were recruited from the Vivantes Klinikum Berlin-Kaulsdorf, Germany, a municipal psychiatry. The clinic staff informed the trained research assistant from the Freie Universität Berlin about patients potentially fitting the inclusion criteria, who were then approached in person. In addition, patients were addressed in the weekly psychoeducation depression group therapy. The healthy control sample was recruited through advertisements in social media, local newspapers and online advertisement platforms and tested by research assistants at the Freie Universität Berlin. In the clinical sample, the presence of a current episode of unipolar depression was required for inclusion. Exclusion criteria for the clinical sample were presence of psychotic symptoms, a bipolar disorder and acute suicidal tendencies. For the control sample, the presence of any mental disorder was an exclusion criterion. For all participants inclusion in the study additionally required a minimum age of 18 years and signed written consent. The inclusion and exclusion criteria were verified by conducting the affective and psychotic disorder modules of the Structured Clinical Interview according to DSM-IV with each participant (SCID; German version:
Upon arrival both at the clinic and at the University laboratory, participants were welcomed and received the informed consent that they were asked to sign. Participants were then interviewed by a trained rater with the SCID. Either directly after the SCID interview or at an appointment shortly afterwards, included subjects completed the intuition task consisting of three blocks (general intuition, fluency, valence; for a detailed description see
The procedure of the JSCT was identical to that of previous studies (
All participants performed three blocks that followed the procedure above but with varying stimulus material (see
After completion of the three intuition blocks, subjects filled out a demographic questionnaire as well as other self-report instruments, not of interest for the current paper (see
Participants’ performance in the JSCT was the main outcome of the study. Trials were discarded in which participants did not provide their coherence judgment within the given time window of 2000 ms. These missed trials were analyzed separately and served us to explore whether patients and healthy controls differed in their ability to react within the given short time window. Solved trials were also discarded from the following coherence judgment analyses because these trials were indicative of explicit insight and not intuition (see
To test H1, we computed a discrimination index for each participant after exclusion of missed responses and solved trials. For this, we first computed hit rates (i.e., the proportion of coherent trials that were correctly judged as coherent, but which were not solved) and false alarm rates (i.e., the proportion of incoherent trials, which had incorrectly been judge as coherent). We then calculated a simple discrimination index by subtracting false alarm rates from hit rates (called Pr in
Hypothesis H2a was tested using a random intercept model, which is conceptually equivalent to fitting a repeated measures ANOVA. In this model, the four conditions of the JSCT fluency block are nested within participants (i.e., each participant contributes four data points, and the random intercept accounts for the fact that the four assessments are usually positively correlated). We used coherence, fluency, and their interaction to predict the percentage of triads that have been judged as coherent (after deleting missed and solved triads). The relevant effect here was the effect of fluency. Note that we only included participants with depression for testing this hypothesis. In contrast, H2b was tested in the full sample, again using a random intercept model. This time, we used coherence, fluency, depression, and their two- and three-way interactions to predict the percentage of triads that have been judged as coherent in the fluency block. The relevant effect here is the interaction effect of depression and fluency. Hypotheses 3a and 3b were tested using the same approach, this time based on data from the JSCT valence block and using valence instead of fluency as a predictor.
In line with the preregistration, we corrected univariate outliers within groups (|z| > 2.5) prior to hypothesis testing using the winsoring method. This way, we corrected 19 data points in 21 variables across 70 participants (1.3%). The criterion for inferences for each hypothesis was
The depressive sample (
Results suggested that depressed patients did not differ significantly from healthy controls regarding the number of missed trials (i.e., trials in which subjects did not respond within the given time window), the number of solved trials (i.e., coherent trials for which the correct solution word was typed in), and the average response time (see
Analyses regarding H1 revealed that depressed patients and healthy control participants did not differ significantly in their ability to discriminate semantic coherence from semantic incoherence in the JSCT general intuition block,
With regard to H2a, results showed that fluency did not significantly predict the percentage of triads that have been judged as coherent in depressed patients,
With regard to H2b, results revealed that the interaction effect of depression and fluency was not significant in predicting coherence judgments,
In line with H3a, analyses revealed that positive valence of word triads significantly predicted semantic coherence judgments in depressed patients,
With respect to H3b, results did not confirm the hypothesis that depression moderated the effect of positive valence on coherence judgments. The effect of positive valence on coherence judgments was not smaller in the depressed sample as compared to the healthy sample,
For testing our main preregistered hypothesis regarding intuition deficits in depressed patients, we selected a simple discrimination index (i.e., the difference between hit and false alarm rates) that has also been used in previous studies (
We also repeated our preparatory analyses and analyses for H1 combining the data from all blocks (i.e., including the data from the fluency and valence block). With regard to H1, again no significant differences emerged, neither for the simple discrimination index nor for A’. We also did not find significant differences in the number of solved trials. However, across all intuition blocks, depressed patients missed more trials and had significantly higher reaction times as compared to healthy controls (see
The aim of the present study was to replicate recently demonstrated deficits in intuitive semantic coherence detection and to explore the effects of processing fluency and positivity on intuition in depressed patients. One major finding was that patients with depression did not differ from healthy controls in their ability to discriminate semantic coherence from semantic incoherence (H1). Even though controls were better at discriminating coherent from incoherent triads compared to depressed patients in the fluency block (indicated by a significant interaction between group and coherence), differences in discrimination indices were not significant when considering data from all blocks.
These findings may query the hypothesis of impaired intuition in depression. However, methodological issues should be considered. The true difference between the groups might be smaller than expected (based on prior research). Thus, our study may have lacked the power to detect it. Moreover, hit and false alarm rates were computed after exclusion of missed trials. Thus, subjects who only responded when they were relatively confident in their judgment (i.e., when seeing a comparably easy triad) might have yielded a higher intuitive discrimination index as compared to subjects who missed relatively few trials. As such, the non-significant difference in intuitive performance might have resulted from depressed patients’ higher tendency to not respond within the given time window, for example when being unsure and less confident, or when confronted with more difficult trials. Indeed, depressed patients missed significantly more trials as compared to healthy control participants when considering their responses across all three intuition blocks.
Our exploratory analyses with reaction times also showed that on average and across all blocks, depressed patients were slower than healthy controls. Together with the finding on missed trials, this result points out that future research would do well in elucidating how longer response times are associated with patients’ intuitive discrimination accuracy. Researchers should hereby distinguish between simple between-subject approaches such as mean reaction time analyses and more sophisticated within-subject methodologies. Using, for example, stochastic diffusion models, can provide important insights into speed-accuracy trade-offs (
Altogether, our findings and considerations call for more replication studies to elucidate the question whether depressed patients are impaired in intuitively detecting meaning and coherence in their environment and within themselves (e.g., meaning in life). Regarding the latter, it seems fruitful to connect intuitive coherence detection research with research on memory coherence, i.e., the ability to construct one’s autobiography in a coherent, integrated way. As memory coherence is associated with psychological health, positive therapy outcomes and seems to buffer protectively against the impact of early life stress (
Results did not reveal any effect of our fluency manipulation, and thus our hypotheses regarding fluency (H2) were not supported. As such, the current study could not replicate previous results that bolstered the fluency model proposed by
The results further showed that positive (vs. negative) valence triggered coherence judgments (H3a) and that this effect was – in contrast to our hypothesis (H3b) – not moderated by depression. This suggests that depressed patients may be susceptible for positive affectivity conveyed by the valence of word triads and used it – when provided externally – in their judgments. This is an important finding, because even though depression is characterized by anhedonia (i.e., the inability to experience positivity), patients seemed to be inclined to detect meaningfulness and coherence when encountering positive valence, bolstering the idea that positivity plays a major role in finding meaning.
How can the current findings be reconciled with previous research depressed patients’ processing of positivity? At first glance they seem to stand in contrast to research showing that – opposed to healthy people – depressed patients do not direct their attention to positivity and are less susceptible to positive stimuli (
Our findings also revealed a significant interaction between valence and coherence in the valence block (i.e., positive word triads that were coherent were most likely to be judged as coherent in both groups). This finding indicates that positivity (conveyed by positive valence in the current study) may lead to more accurate intuitive judgments and is in line with previous research showing that not only “tonic” positive affect (e.g., manipulated or freestanding positive mood;
Along this line, it is of important practical relevance to test whether depressed patients can themselves produce the positive affect needed to go with their intuition in daily life. Extending laboratory research, a recent daily diary study found that people are not only inclined to make decisions intuitively when they are in a good mood (as compared to a negative mood,
From a therapeutic perspective, the current findings imply that targeting positive affect in psychotherapy may be important in fostering patients’ ability to detect coherence. It would be an important next step to investigate the intuitive detection of meaning not only with regard to laboratory stimuli but also on a broader level with regard to finding coherence and meaning in life. Hereby, clinical researchers may build upon recent basic psychological research on how intuitive processing, positive affect, and finding meaning in life interact (
A number of limitations should be taken into account. First, even though the sample size was in compliance with the a-priori power analysis, it was still relatively small. Thus, future studies should test our assumptions with larger samples to increase the power and reliability of findings. Furthermore, a limitation of the current study was that the samples were not matched in terms of educational level. Even though the relatively lower educational level of depressed patients is consistent with epidemiological studies showing that the prevalence of psychological disorders is higher in low socioeconomic groups, future studies should take care of the matching issue to avoid potential confounds. In addition, we randomized different factors such as the key position for the coherence judgments and the stimuli that were either presented in the main intuition block or the fluency bock. Even though randomization is of methodological importance, this may have led to reduced comparability of responses between subjects and thus to reduced power to detect group differences. Thus, future research should use larger sets of stimuli and larger sample sizes in order to ensure randomization and reduce statistical noise. Furthermore, conclusions with regard to the role of positivity should be drawn cautiously because our study was lacking a neutral control condition. Thus, we cannot rule out that, for example, reduced negativity (as opposed to increased positivity) drove our effects in the valence block. Future studies should test whether positivity (e.g., conveyed by the valence of stimuli) alters subtle affective responses in subjects. Only by this means we can conclude whether positive affect elicited in subjects triggers coherence judgments (see
Albeit these considerations, the current study presents an important contribution to the field. It is a preregistered replication study which follows current state-of-the-art demands to bolster the robustness of psychological research findings. In addition, we used experimental paradigms from basic psychology and hereby build the bridge from basic to clinical research. Altogether one may conclude from the current study that the cognitive profile of depressed patients is not merely deficient. The results elucidate the importance of positivity when it comes to detecting meaning and coherence. The latter is of major clinical importance, because in a depressed state, people often experience their life as meaningless and cannot find coherence. Whether promoting positivity may not only enhance how patients feel but will also help them to find meaningfulness and to follow their intuitions is a fruitful endeavor to study for future research.
The supplementary materials include the preregistration protocol for this study (for access see
Where possible, we based our power analyses on the effect sizes found in earlier work. In particular, the effect size of the intuition impairment in patients with depression compared to healthy controls was
Variable | Patients |
Controls |
|||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
General Intuition Block | |||||||||||
Discrimination Index | 0.18 | 0.17 | 0.19 | 0.18 | 0.00 | -0.12 | 68 | 0.90 | -0.09 | 0.08 | -0.03 |
A‘ | 0.65 | 0.13 | 0.65 | 0.14 | 0.00 | -0.02 | 68 | 0.98 | -0.06 | 0.06 | -0.01 |
Number of missed trials | 7.17 | 4.73 | 5.17 | 4.23 | -2.00 | 1.86 | 68 | 0.07 | -0.14 | 4.14 | 0.45 |
Number of solved trials | 0.63 | 0.91 | 0.89 | 1.05 | 0.26 | -1.09 | 68 | 0.28 | -0.73 | 0.21 | -0.26 |
Average reaction time per trial (in seconds) | 1.11 | 0.15 | 1.04 | 0.20 | -0.08 | 1.79 | 68 | 0.08 | -0.01 | 0.16 | 0.43 |
Combined blocks | |||||||||||
Discrimination Index | 0.17 | 0.11 | 0.20 | 0.10 | 0.03 | -0.98 | 68 | 0.33 | -0.08 | 0.03 | -0.24 |
A‘ | 0.64 | 0.09 | 0.66 | 0.09 | 0.02 | -1.07 | 68 | 0.29 | -0.06 | 0.02 | -0.26 |
Number of missed trials | 15.90 | 10.76 | 10.85 | 8.79 | -5.05 | 2.15 | 68 | 0.04 | 0.36 | 9.74 | 0.52 |
Number of solved trials | 1.87 | 2.47 | 2.19 | 2.26 | 0.31 | -0.55 | 68 | 0.58 | -1.45 | 0.82 | -0.13 |
Average reaction time per trial (in seconds) | 1.02 | 0.20 | 0.93 | 0.19 | -0.10 | 2.11 | 68 | 0.04 | 0.01 | 0.19 | 0.51 |
We thank Lola Hermann, Isabelle Klausener, and Leonard Wegner for their help in data collection.
This research was facilitated by research funds of the Forschungskommission of the Freie Universität Berlin.
The authors have declared that no competing interests exist.