As shown in , the RT analysis revealed significantly slower overall responses times for patients than for controls, F (1, 58) = 12.33, p=.001. However, group differences in overall accuracy approached but did not reach significance F (1, 58) = 2.99, p=.09. In addition, both patients and controls showed a similar pattern of accuracy across cue and flanker conditions, as reflected by the absence of significant interaction of either of these factors with group.
Reaction and Accuracy as a Function of Experiment Condition for Patient and Control Groups
Ratio network measures were next submitted to a mixed analysis of variance with group (patients, controls) as a between-subject factor and network score (alerting, orienting, executive control) as a within-subject factor. The group main effect, F (1, 58) = 12.90, p = .001, revealed lower ratio scores across alertness, orienting, and executive control network measures for the patients in comparison to the controls (see ). Most important is that network score interacted significantly with group, F (2,116) = 8.50, p < .001. As shown in , the patient group showed a pronounced reduction for alertness, t (58) = 2.25, p < .05, but a very similar level of visual orienting to the control group, nor did the group differ significantly in executive control.
Attention Network scores (in RT and ratio score,) mean RT, and accuracy for patients with schizophrenia and controls.
Network Ratio Scores for Patient and Control Groups
To compare group patterns of RT as a function of cue and flanker, we ran an analysis of covariance (covariate: overall RT) with group (patients, controls) as a between-subject factor and cue (no cue, double cue, center cue, spatial cue) and flanker (neutral, congruent, incongruent) as within-subject factors. As expected, both groups responded fastest to targets preceded by spatial cues, F (3,171) = 11.16, p < .001. Of special relevance is that flanker interacted with group, F (2, 114) = 6.27, p < .01. While both groups showed a similar pattern of response times for congruent and neutral trials, they showed different patterns of response times for the comparisons of incongruent and congruent trials, F (1, 57) = 5.92, p<.05, and incongruent and neutral trials, F (1, 57) = 7.51, p<.01. In both instances, in relation to the control group, the patient group, unexpectedly, showed reduced interference (incongruent-congruent) and reduced cost in response times for targets surrounded by incongruent flanker (incongruent-neutral).
Also of note is that while group interacted with cue, F (3, 171) = 2.612, p<.05, collapsing over flanker type did not reveal the source of this significant interaction of cue and group. However, in comparison to controls, the patient group consistently showed slower response times for targets surrounded by neutral flankers regardless as to whether the target was preceded by either a no cue, F (2, 114) = 3.08, p=.05, a double cue, F (2, 114) = 7.34, p = .001, or a center cue, F (2, 114) = 3.31, p=.05, with the notable exception for targets surrounded by neutral flankers preceded by a spatial cue. In other words, the patients showed a disproportionately slowing in RT to targets surrounded by non-informative neutral flankers across three of the four levels of the cue factor --- that is for targets preceded by either no cues, double cues, or center cues, but not for spatially cued-targets.
Spearman rank correlations revealed for the general patient group, lower levels of ANT alertness correlated with longer illness duration rho (27) = −.503, p<.01 and higher medication levels rho (26) = −575, p <.01. Slower overall RT on the ANT for the patient group also correlated with higher medication levels rho (26) = .579, p<.01. For the patient subset, shows that slower overall RT on the ANT correlated significantly with lower volume for left, rho (18) = −.709, p=.001, and right, rho (18) = −.779, p<.001, CB. However, only the correlation of slower overall RT on the ANT and right CB volume remained significant when controlling for medication level, as reflected by the partial correlation value of −.572, p<.05 or after controlling for illness duration, as reflected by the partial correlation value of −.582, p<.05. In addition, reduced alerting efficiency correlated significantly with lower right CB volume, rho (18) = .494, p<.05, but this correlation only approached significance after controlling for medication, as reflected by the partial correlation value of .457, p=.075, and was no longer significant when controlling for illness duration (p>.35). Reduced visual orienting correlated significantly with lower left CB fractional anisotropy, rho (18) =.484, p<.05, and this correlation remained significant when controlling for medication level, as reflected by the partial correlation value of .495, p<.05 and for illness duration, as reflected by the partial correlation value of .759, p<.01.
Spearman rank correlations of cingulum bundle and ANT measures for the patient subset.