روابط منحصر به فرد از سن و بزهکاری با کنترل شناختی

Abstract Context processing has significant empirical support as an explanation of age- and psychopathology-related deficiencies in cognitive control. We examined whether context processing generalizes to younger individuals who are in trouble with the law. We tested whether age and delinquency might have unique relations to context processing skills in four samples of male participants: adolescent offenders (n = 43), control adolescents (n = 33), young adult offenders (n = 40), and control young adults (n = 31). We used a modified Stroop task to measure context processing (i.e., attention, memory, and response inhibition). Task performance was superior for older participants in conditions most demanding of context processing skills. Adolescent offenders and control adolescents showed difficulties engaging selective attention to filter out irrelevant information, even after controlling for the effects of age. Control adolescents made the most errors in the condition most demanding of context processing, whereas the other three samples showed slower processing but fewer errors in context processing.

Results We conducted a mixed analysis of variance (ANOVA) to examine the effects of task, delay, condition, age, and all possible interactions among these variables. We also included the main effect of sample as well as all possible interactions of sample with task, delay, and condition. Because we include both age and sample in the same statistical model, tests of the effects of age and sample examine the unique predictive ability of each to explain performance on the Stroop task. When interpreting tests of these two variables in the same model, we must focus on the effects of age that are unrelated to sample and the differences between samples that are unrelated to age because variability that they share will be collinear and not influence the tests of the individual effects (Cohen, Cohen, West, & Aiken, 2003). We therefore interpret the effects of age as primarily representing biological development and the effects of sample as primarily representing psychosocial characteristics. Post hoc comparisons were conducted for all significant effects using α = 0.05 with a Sidak (1967) correction for multiple comparisons. Because participant characteristics could have different effects at different combinations of our factors, we examined the most likely individual difference variables between our samples that could influence task performance, including: ethnicity, socioeconomic status, number of head traumas, drug and alcohol use, composite IQ, distractions, and diagnosis. We performed one multivariate analysis of variance (MANOVA) for each characteristic examining the collective influence of the covariate on the error rates broken down by task, condition, and delay. After controlling for experiment wise error rates using a Sidak correction (p < 0.007), none of the participant characteristics had a significant influence on performance. Any potentially significant findings based on sample cannot therefore be attributed to differences among samples on these characteristics. Error rates – main effects The ANOVA table for the effects on error rates is presented in Table 2. There was a significant main effect of task, such that the mean error rate for color naming (Mraw = 0.12, SEraw = 0.007) was significantly larger than that for word reading (Mraw = 0.05, SEraw = 0.004). There was a significant main effect of condition, such that the mean error rate for incongruent trials (Mraw = 0.45, SEraw = 0.021) was significantly larger than that for neutral (Mraw = 0.04, SEraw = 0.008) and congruent trials (Mraw = 0.04, SEraw = 0.008). Relative to the neutral condition, participants were less accurate when the color word and ink color mismatched, but were not more accurate when the color word and ink color matched. This suggests the presence of an interference effect (i.e., when incongruency between ink colors and color words leads to slower reaction times) but the absence of a facilitation effect (i.e., when congruency between ink colors and color words leads to faster reaction times). Table 2. ANOVA model predicting arcsine transformed error rates. Effect Wilks’ lambda Test statistic p-value Condition 0.25 F(2, 138) = 207.00 <0.001 Task 0.49 F(1, 139) = 146.30 <0.001 Delay 0.99 F(1, 139) = 1.58 0.21 Age – F(1, 139) = 1.46 0.23 Sample – F(3, 139) = 1.30 0.28 Condition × task 0.49 F(2, 138) = 71.72 <0.001 Condition × delay 0.90 F(2, 138) = 7.96 0.001 Condition × age 0.95 F(2, 138) = 3.31 0.04 Condition × sample 0.88 F(6, 276) = 3.10 0.01 Task × delay 0.99 F(1, 139) = 1.48 0.23 Task × age >0.99 F(1, 139) = 0.19 0.67 Task × sample 0.94 F(3, 139) = 2.94 0.04 Delay × age >0.99 F(1, 139) = 0.64 0.43 Delay × sample 0.97 F(3, 139) = 1.42 0.24 Condition × task × delay >0.99 F(2, 138) = 0.15 0.86 Condition × task × age 0.99 F(2, 138) = 0.57 0.57 Condition × task × sample 0.96 F(6, 276) = 1.04 0.40 Condition × delay × age >0.99 F(2, 138) = 0.11 0.90 Condition × delay × sample 0.92 F(6, 276) = 1.86 0.09 Task × delay × age 0.99 F(1, 139) = 1.69 0.20 Task × delay × sample 0.96 F(3, 139) = 1.89 0.14 Condition × task × delay × age 0.95 F(2, 138) = 3.63 0.03 Condition × task × delay × sample 0.89 F(6, 276) = 2.69 0.02 Note: Wilks’ Lambda not calculated for between-subjects effects. Table options Error rates – task-related interaction effects The significant task by condition interaction indicated that in the incongruent condition, mean error rates for color naming (Mraw = 0.31, SEraw = 0.01) were significantly larger than those for word reading (Mraw = 0.11, SEraw = 0.004). This suggests that interference effects were stronger in the color naming task than in the word reading task. The significant delay by condition interaction indicated that in the neutral condition, mean error rates for the short delay (Mraw = 0.03, SEraw = 0.003) were larger than those in the long delay (Mraw = 0.01, SEraw = 0.001). However, the reverse was true in the incongruent condition. Error rates – sample and age-related interaction effects The significant condition by age interaction indicated that age did not appear to have substantial effects in the neutral or congruent conditions. In the incongruent condition, however, older participants had lower error rates than younger participants. The significant condition by sample interaction indicated that mean error rates were not significantly different between samples in the congruent and neutral conditions. However, in the incongruent condition, control adolescents had the highest error rates (Mraw = 0.26, SEraw = 0.02), followed by the young adult offenders (Mraw = 0.22, SEraw = 0.02), adolescent offenders (Mraw = 0.20, SEraw = 0.01), and control young adults (Mraw = 0.18, SEraw = 0.02). The significant task by sample interaction indicated that across all four samples, mean error rates for the color naming task were higher than those for the word reading task. For all of the samples, the difference between color naming and word reading was significant. However, the difference between the color naming and word reading for control adolescents was notably larger than that for any of the other three samples. The four–way interaction between condition, task, delay, and age was significant (see Fig. 1). The largest differences in this interaction are observed in the incongruent condition. Substantial age effects are visible for incongruent stimuli in the short delay condition of the color naming task, the long delay condition of the color naming task, and the long delay condition of the word reading task. There does not appear to be an effect in the short delay condition of the word reading task. Four-way interaction between age, task, condition, and delay. Fig. 1. Four-way interaction between age, task, condition, and delay. Figure options The four–way interaction between task, delay, condition, and sample was significant (See Fig. 2). Across all combinations of delay and sample, there were no significant differences between mean error rates in the congruent and neutral conditions. Mean error rates in the neutral and congruent conditions also did not vary by task in any of the samples. The effects of sample, task, and delay were only observed in the incongruent condition. For the control young adults and the young adult offenders, the difference between color naming and word reading tasks in the incongruent condition was larger in the short delay than in the long delay condition. For the adolescent offenders, the difference between tasks in the incongruent condition was approximately the same between the short and long delay conditions. For the control adolescents, the difference between tasks in the incongruent condition was smaller in the short delay than in the long delay condition. Four-way interaction between sample, task, condition, and delay. Fig. 2. Four-way interaction between sample, task, condition, and delay. Figure options Reaction times Although results from reaction times did not always reach the significance level of error rates, the patterns for reaction times across conditions and tasks were similar to those for error rates. There may have been more variability in reaction times because reaction times, but not error rates, on the Stroop task have been related to reading disabilities (Protopapas, Archonti, & Skaloumbakas, 2007), which are fairly common in incarcerated samples. The one important difference between the results for reaction times and the results for error rates was the presence of a significant main effect of sample on reaction times (F[3134] = 3.93, p = 0.01; See Fig. 3). Whereas the error rates were approximately the same across all four samples, community youth had faster reaction times than the other three samples (which were not significantly different from each other). The presence of higher error rates among control adolescents may be indicative of a speed-accuracy trade-off, although this is only weakly supported since there was not a significant effect of sample on error rates. Comparing error rates and reaction times by sample. Fig. 3. Comparing error rates and reaction times by sample.