Catecholaminergic Modulation of Semantic Processing in Sentence Comprehension

Abstract

Catecholamine (CA) function has been widely implicated in cognitive functions that are tied to the prefrontal cortex and striatal areas. The present study investigated the effects of methylphenidate, which is a CA agonist, on the electroencephalogram (EEG) response related to semantic processing using a double-blind, placebo-controlled, randomized, crossover, within-subject design. Forty-eight healthy participants read semantically congruent or incongruent sentences after receiving 20-mg methylphenidate or a placebo while their brain activity was monitored with EEG. To probe whether the catecholaminergic modulation is task-dependent, in one condition participants had to focus on comprehending the sentences, while in the other condition, they only had to attend to the font size of the sentence. The results demonstrate that methylphenidate has a task-dependent effect on semantic processing. Compared to placebo, when semantic processing was task-irrelevant, methylphenidate enhanced the detection of semantic incongruence as indexed by a larger N400 amplitude in the incongruent sentences; when semantic processing was task-relevant, methylphenidate induced a larger N400 amplitude in the semantically congruent condition, which was followed by a larger late positive complex effect. These results suggest that CA-related neurotransmitters influence language processing, possibly through the projections between the prefrontal cortex and the striatum, which contain many CA receptors.

Keywords: N400; catecholamine; language processing; late positive complex (LPC); methylphenidate.

Figure 1

Mean response accuracy (d′) and RTs for each condition in Order 1 and Order 2 separately. Order 1 took the MPH-placebo order, while Order 2 took the reversed order. “Con” represents congruent condition and “Inc” represents incongruent condition. The error bars represent corrected standard error of the mean (Cousineau 2005). For illustration purposes, the figure uses raw RTs for subjects’ performance in each condition, despite the fact that all analyses were conducted using log-transformed RTs.

Figure 2

Grand-averaged ERPs (n = 42) in the Semantic task. (A) Waveforms at nine representative electrodes timed-locked to the critical nouns in the semantically congruent (CON) versus incongruent (INCON) and MPH versus placebo (PLA) conditions. The negativity is plotted upward. For illustrative purpose only, a 15-Hz low-pass filter has been applied on the waveforms. (B) Scalp distributions of the semantically congruent and incongruent conditions on MPH and placebo. The electrodes that were included in the significant cluster of MPH × Congruence interaction were plotted as well. The positive interactions suggested that the mean amplitude difference between congruent and incongruent (INC–CON) conditions was smaller in the MPH than the placebo condition, and the negative interaction suggested an opposite direction (^*P < 0.01, ×P < 0.05). (C) Mean amplitudes in the electrodes that showed significant interaction effect in each time window. Consistent with ERPs in (A), the negativity is plotted upward as well. For the time windows that showed no significant interactions (i.e., 350–500 and 900–1200 ms), we plotted the mean amplitudes for the same electrodes as in the earlier time window (i.e., 250–350 and 550–900 ms, respectively), in order to show the continuation of the previous effect. The error bars represent corrected standard error of the mean for a within-subject design. As discussed in the Methods section, the observation of significant clusters did not provide information on the exact spatial extent.

Figure 3

Grand-averaged ERPs (n = 42) in the Font-size task. (A) Waveforms at nine representative electrodes timed-locked to the critical nouns in the semantically congruent (CON) versus incongruent (INCON) and MPH versus placebo (PLA) conditions. The negativity is plotted upward. For illustrative purpose only, a 15-Hz low-pass filter has been applied on the waveforms. (B) Scalp distributions of the semantically congruent and incongruent conditions on MPH and on placebo. The electrodes that were included in the significant cluster of MPH × Congruence interaction were plotted as well. The negative interaction suggested that the mean amplitude difference between congruent and incongruent (INC–CON) conditions was larger in the MPH than the placebo condition (^*P < 0.01, ×P < 0.05). (C) Mean amplitudes in the electrodes that showed significant interaction effect in each time window. Consistent with ERPs in (A), the negativity is plotted upward as well. For the time windows that showed no significant effect (i.e., 100–150, 550–900, and 900–1200 ms) in the Font-size task, we plotted the mean amplitudes for the same electrodes in each window as in the Semantic task, in order to allow cross-task comparisons. The error bars represent corrected standard error of the mean for a within-subject design.

Figure 4

Grand-averaged amplitude differences between the semantically congruent (CON) and incongruent (INCON) sentences in the MPH and placebo (PLA) conditions in the Semantic (Sem) and Font-size (Font) tasks. The negativity is plotted upward.

Figure 5

Hypothesized neurotransmitter mechanisms of the catecholaminergic system’s gating function. The figure is adapted from Figure 4 in Cools (2019). We hypothesized that the CAergic system could up- and downregulate language processing by modifying the striatum to prefrontal cortex (PFC) projections. An increased striatum CA level provides a relevance signal for the PFC to amplify the saliency of language input, and thus supports the semantic combinatorial processing within the PFC.