Trait-like Features of Rumination

To examine whether differential performance on the tasks is due to personality traits such as rumination, we may address the question of whether the same people who show biases on the lexical decision task also show biased performance on the valence identification task. For this analysis, the relative delay on negative words on the lexical decision task may be correlated with the relative facilitation for negative words (or delay for positive words) on the valence-identification task, regardless of whether research participants are depressed, as determined by the administered mood measures. Similarly, relative delay on the valence-identification task may be used as an independent variable along with depression on a regression similar to the one described above. Interactions of depression and rumination in forming interference on the task may be captured by adding these terms specifically into the analysis.

Toward this end, hierarchical multiple regressions were performed using variables representing the difference in reaction times to negative vs neutral, positive vs neutral, and negative vs positive words on the lexical decision task, entering BDI on the first step of the regression, and variables representing the analogous differences on the valence identification task on the second step. In this way, the unique contribution of valence-mediated reaction-time biases above and beyond depression could be examined.

Together BDI score, the valence identification differential reaction times, and their interactions accounted for 11$\%$ of the variation in the negative vs. neutral contrast for the lexical decision task. The interaction of valence-identification task performance and depression (a measure of the proposed ``ruminative coping style'' for depression) accounted for 3.9$\%$ of the variation in lexical decision task performance, more than was accounted for by depression or the valence identification task alone, but not enough variation to reach statistical significance, $F_\delta$(3,39)=.864, p=.43. A one millisecond facilitation in the negative vs. neutral contrast on the valence-identification task is associated with a -.185ms delay on the negative vs. neutral contrast for the lexical decision task, t(39)=-1.03, p=.27, and the a 1ms facilitation on the positive vs. negative contrast for the valence identification task is associated with a .27ms facilitation on the negative vs. neutral contrast on the lexical decision task, t(39)=1.03, p=.30. These results are expected if the lexical decision task is measuring an interference process while they are not expected if both tasks measure resources devoted to valence identification. Together the variables associated with biased performance on the valence identification task accounted for 7.4$\%$ of the variation in the difference between reaction times to negative and neutral words, above and beyond the .002$\%$ of variation accounted for by BDI alone, $F_\delta$(3,41)=1.66, p=.202.

Results are even stronger for the positive vs. neutral contrast. Together BDI score, the valence identification differential reaction times, and their interactions accounted for 31$\%$ of the variation in the positive vs. neutral contrast for the lexical decision task, F(5,39)=3.5, p=.01. Again, the interaction of depression and valence identification task reaction times was not statistically significant, $R_\delta^2=.03$, $F_\delta$(3,39)=.75, p=.47. The negative vs. neutral contrast for performance on the valence identification task was associated with -.25ms delays on the lexical decision task, t(39)=.526, p=.135, and the positive vs. neutral contrast on the valence identification task was associated with a .21ms facilitation, t(39)=.803, p=.42. The contribution of individual contrasts was not statistically significant due to the large amount of inter-subject variability. Still, valence-identification performance accounted for 28.4$\%$ of the variation in lexical decision performance on the positive vs. neutral contrast, $F_\delta$(3,41)=5.42, p=.003, with depression alone accounting for only .03$\%$ of the variation, F(1,43)=1.43, p=.21.

Results are similar for the positive vs. negative contrast. Together BDI score, the valence identification differential reaction times, and their interactions accounted for 22.5$\%$ of the variation in the positive vs. negative contrast for the lexical decision task, F(5,39)=2.26, p=.06. The interaction terms accounted for only 2$\%$ of the variation, $F_\delta$(2,39)=.47, p=.62, with the bulk of the variation being accounted for by the valence-identification task performance $R_\delta^2=.18$, $F_\delta$(3,39)=4.18,p=.01, and BDI alone accounting for only 2$\%$ of the variation in the differential lexical decision task performance, F(1,41)=.86, p=.35.

Thus, it appears as if biased performance on the lexical decision task is primarily a function of rumination, as measured by facilitation on negative words, and delays in responding to positive words on the valence identification task. Depression alone does not account for a great deal of the observed delays on this task, and the interaction of depression and valence identification performance, the operationalization of rumination on depressive constructs, while accounting for more variation than depression alone, also does not account for a statistically significant amount of variation.