Patients were consecutively recruited from our Neurological Department, including 32 non-demented ALS patients (21 males, 11 females; mean age, 57 years) and 10 non-demented PLS patients (one male, nine females; mean age, 52 years; Table 1). The clinical diagnosis was based on the El Escorial revised criteria and included patients with laboratory-supported probable, probable, and definite ALS (Brooks et al., 2000). PLS was diagnosed according to Pringle’s criteria (Pringle et al., 1992). Patients with a current or past history of neurologic disease other than MND, with other neuropsychiatric or medical disorders, or who used psychoactive drugs within the prior 2 months were excluded. Clinically healthy control subjects had a similar age and sex distribution (12 males, 15 females; mean age, 57 years), a normal neurologic examination, and no history of neurologic or psychiatric disorders. The participants provided informed consent to participate in the study, which was approved by the local ethics committee.

Demographic data and detailed neurological findings were recorded. Disease severity was estimated using the ALS Functional Rating Scale (ALS-FRS; The ALS CNTF Treatment Study Phase I-II Study Group, 1996). The patients were graded in terms of upper motor neuron (UMN) “burden” by totaling the number of pathological UMN signs upon examination. These were taken as pathologically brisk biceps, supinator, triceps, finger, knee, and ankle reflexes, and extensor plantar responses assessed bilaterally and brisk facial and jaw jerks (range: 0–16; Turner et al., 2004). Disease duration at the time of the investigation was calculated in months, from the date of first symptoms to the date of examination. The rate of disease progression was calculated using the following formula (Ciccarelli et al., 2006): disease progression rate = (40 – ALS-FRS score)/disease duration.

Muscle strength was graded on the Medical Research Council (MRC) scale, from 0 to 5, in selected upper and lower limb (LL) muscle groups. On each side, seven upper limb (UL) muscles and five LL muscles were tested as previously described (de Carvalho et al., 2003). Three scores were obtained: UL-MRC sum score (maximum: 70), LL-MRC sum score (maximum: 50), and total MRC sum score (UL + LL MN score; maximum: 120).

Patients with other neuropsychiatric or medical disorders or who used psychoactive drugs during the previous 2 months were excluded.

Reaction times (RTs) in the Stroop task were measured using a computerized version implemented in commercial STIM software (Neuroscan, Herndon, VA, USA). Responses were recorded using a computer mouse with two response buttons. Four color words (green, red, yellow, and blue) written in congruent (50%) or incongruent (50%) color were randomly presented (stimulus duration, 200 ms; intertrial interval, 3.5 s) in four different series of 32 stimuli each.

In the first condition (simple RT – SRT), the subjects had to press a button for every stimulus presentation, regardless of stimulus type. The second condition (go/no-go RT) consisted of two series, in which a response was required to either the incongruent (go/no-go I) or congruent (go/no-go C) stimuli. In the third condition (choice RT), the subjects had to press one button after the congruent stimuli (choice C) and the other button after the incongruent stimuli (choice I). For each series, the response latency was measured only for correct responses. Trials with latencies that exceeded 2.3 s were considered omissions and excluded from the calculation of average RTs and accuracy. The latter was calculated in the complex RTs (go/no-go and choice) as the percentage of correct responses.

Twenty-nine electroencephalography (EEG) channels with binaural reference were recorded using scalp electrodes set on an elastic cap (Electrocap International, Eaton, OH, USA). The EEG signal was amplified (Synamps, Neuroscan, Herndon, VA, USA), filtered (DC–50 Hz), and digitized (sampling frequency, 250 Hz). The electrooculogram and electromyogram of the right and left extensor pollicis brevis were also recorded to detect eye movements and relaxation failure.

A series of 120 of the same Stroop stimuli used for the RT measurement were presented (stimulus duration, 200 ms; intertrial interval, 6 s). The subjects were instructed to mentally discriminate between congruent and incongruent stimuli. This condition was chosen for ERP recording to avoid movement interference. Attention was monitored by randomly asking the subjects in every 10–15 trials to verbally define the congruency of the last stimulus presented.

Epochs from −500 to 1200 ms from stimulus onset were obtained. Linear detrending was performed over the entire epoch to correct for DC drifts. The baseline was then corrected between −500 and 0 ms. Epochs that contained artifacts or muscle relaxation failure upon visual inspection were excluded from the analysis. Initially, separate averages were obtained for congruent and incongruent stimuli. After a preliminary comparison between and within groups, which did not show significant differences between the parameters obtained in the two conditions, data from the congruent and incongruent trials were collapsed into a single ERP for each subject to reduce signal noise.

The latency of the main ERP components [i.e., N1 (O1 or O2 electrode), P2, N2, P3, N4, and LPC (late positive complex, peaking 600–700 ms post-stimulus onset; Fz electrode)] was measured for each subject. The amplitude and topographic analysis was performed at time intervals of the same components (time intervals = group mean latency value of each component ±30 ms) using low-resolution brain electromagnetic tomography (LORETA; Pascual-Marqui et al., 1994; Pascual-Marqui, 1999; see Statistical Analysis below).

The significance of group effects with regard to the number of correct responses (in the choice condition, go/no-go C condition, and go/no-go I condition), RT latency in the choice C condition, choice I condition, go/no-go C condition, go/no-go I condition, and simple RT condition, and latency of the main ERP components (N1, P2, N2, P3, N4, and LPC) was tested using three separate multivariate analyses of variance (MANOVAs). Post hoc tests were performed using Bonferroni correction. Correlations between clinical scales and RTs and between clinical scales and ERP latencies were also performed using Spearman’s coefficient for interval scales and Pearson’s coefficient for ordinal scales. All of the statistical tests were performed using SPSS 17 software (Technologies, Chicago, IL, USA). Group differences in the amplitude and topography of ERP waveforms were investigated using LORETA with a statistical non-parametric voxel-wise comparison between the ALS, PLS, and control groups. The level of significance was set at p < 0.05.

In the ALS patient group, the mean scores were the following: mean UMN score: 9; mean ALS∖FRS rate: 32.5; mean bulbar score: 11.03; mean MRC total score: 101.1; mean disease progression rate: 0.64. In the PLS patient group, the mean scores were the following: mean UMN score: 14; mean ALS∖FRS rate: 33.5; mean bulbar score: 10.5; mean MRC total score: 117.5; mean disease progression rate: 0.26 (Table 1).

The MANOVA revealed significant group effects in the percentage of correct responses (F = 5.057, p = 0.003). In the go/no-go I condition, ALS patients committed significantly more errors than controls (p = 0.014) and PLS patients (p = 0.05). No significant difference was found between PLS patients and controls (Figure 1).

The MANOVA also revealed significant group effects in RTs (F = 5.437, p < 0.001). In fact, PLS patients appeared significantly slower than controls (p < 0.001) and ALS patients (p = 0.002) in the SRT condition. ALS patients instead showed no significant differences in RTs compared with controls (Figure 2). RT was negatively correlated with bulbar score in the go/no-go I condition (ρ = −0.455, p = 0.008) and disease progression rate in the simple RT condition (ρ = −0.357, p = 0.038).

The MANOVA revealed a significant group effect in ERP latencies (F = 3.627, p = 0.004). N1, N2, P3, and LPC latencies were significantly delayed in ALS patients compared with controls (N1: p = 0.021; N2: p = 0.043; P3: p = 0.036; LPC: p = 0.004). No latency differences were found between PLS patients and controls or between ALS and PLS patients (Figure 3). A positive correlation was found between N4 latency and disease duration (r = 0.446, p = 0.006).

The LORETA non-parametric voxel-wise analysis revealed significant group differences. In the P2 time window (Figure 4), the ALS group exhibited significantly decreased activation of the left superior and middle temporal gyri compared with controls. In the P3 (Figure 5) and N4 (Figure 6) time windows, ALS patients exhibited significantly reduced activation of the ACC and medial frontal gyrus compared with controls. No differences were found between PLS patients and controls.

The reduced RT accuracy found in ALS patients, who committed significantly more errors than the other two groups, suggests that conflict-monitoring function, which is necessary for processing competing information and is supposedly mediated by the ACC (Botvinick et al., 1999, 2001), was impaired in ALS patients. Our RT findings are consistent with previous studies that reported executive function deficits in ALS (Abrahams et al., 1997, 2000; Goldstein et al., 2011; Phukan et al., 2012).

The RT results in PLS patients were not consistent with previous neuropsychological studies, which reported impaired performance in PLS patients in a subgroup of tasks used for neuropsychological assessment. Functional frontal lobe impairments appeared to be similar to cognitive dysfunction in ALS (Caselli et al., 1995; Piquard et al., 2006; Grace et al., 2011). In the present study, PLS patients were significantly more accurate than ALS patients, and their RT motor performance was slower, which was expected when considering motor symptoms, but qualitatively similar to healthy controls in terms of accuracy. However, in these previous studies, the subjects had to emit a vocal or motor response, and the results cannot be considered completely independent from movement deficits. Therefore, motor impairments may at least partially account for the inconsistency with our present results.

Amyotrophic lateral sclerosis patients presented significant delays in several Stroop ERP components compared with controls, mainly in the N1, N2, P3, and LPC components. N1 is assumed to reflect selective attention to basic stimulus characteristics, initial selection for later pattern recognition, and intentional discrimination processing (Vogel and Luck, 2000). Its source is located in the inferior occipital lobe, occipito-temporal junction (Hopf et al., 2002), and inferior temporal lobe (Bokura et al., 2001). The N2 component in go/no-go-like tasks has been attributed to response inhibition mechanisms (Jodo and Kayama, 1992; Gonzalez-Rosa et al., 2013). However, the N2 component has also been reported to occur in relation to covert responses in the present study and previous studies (Pfefferbaum et al., 1985). This would indicate that it is not completely attributable to the inhibition of responses and that it may at least partially account for conflict monitoring. N2 is especially pronounced over the fronto-central electrodes and has been proposed to reflect ACC sensitivity to conflict (van Veen and Carter, 2002).

The P3 component is elicited in tasks related to stimulus differentiation and appears when a memory representation of the recent stimulus context is updated upon the detection of deviance from it (Sutton et al., 1965). The frontal P300 component in go/no-go-like tasks has been associated with an inhibitory mechanism (Gajewski and Falkenstein, 2013). However, in the present study, the subjects only had to mentally discriminate between congruent and incongruent stimuli; therefore, conflict did not arise at the response level. Thus, the P3 component observed herein most likely reflects the detection of conflict that arose at the level of the semantic encode.

The LPC component is extended positivity that peaks 600–700 ms after stimulus onset and is attributable to the semantic processing of word meaning (Liotti et al., 2000). The ERP delay observed in the ALS group is consistent with previous studies that used oddball paradigms (Hanagasi et al., 2002; Paulus et al., 2002; Ogawa et al., 2009). Altogether, our RT and ERP findings reflect cognitive control impairment in ALS patients.

Primary lateral sclerosis patients showed no abnormalities in ERP latency, despite the important slowing of RTs. This result is only apparently in opposition to the RT results, in which lower accuracy without slowing was observed in ALS patients, and a delay but normal accuracy was observed in PLS patients. In fact, slower RTs in tasks that require a motor response in PLS patients appear to be mainly related to movement impairments, as suggested by their normal frequency of correct responses.

Moreover, the lack of cognitive processing abnormalities in PLS, particularly with respect to ALS, unlikely depends on differences in disease duration because this feature would have favored the ALS group in our sample.

Clinical scores that assess upper and lower motor neuron impairment were correlated with RTs but not with ERP latencies, which were delayed in ALS. This finding allows us to disentangle cognitive and movement impairments that contribute to task performance. Mental discrimination of the Stroop stimuli accounted for cognitive processing, independent of motor performance, suggesting that this version of the task may be useful for cognitively assessing patients in advanced stages of the disease when profound motor disability interferes with communication or leads to a virtual locked-in syndrome. Furthermore, this approach may be useful for evaluating eligibility for brain-computer interface. Finally, computerized RTs, which provide useful measures in the assessment of executive function, may reflect the severity of motor impairment as indicated by their correlation with clinical assessment.

In ALS patients compared with controls, the voxel-wise group comparison of ERPs using LORETA showed a significant decrease in the activation of the left superior and middle temporal gyri in the P2 time window, suggesting impaired lexical processing, and a significant decrease in activity in the ACC and medial frontal gyrus in the P3 and N4 time windows, suggesting impaired conflict monitoring.

The findings of decreased activity in our sample of ALS patients are consistent with previous ERP studies that reported hypoactivity in frontotemporal regions in ALS patients (Vieregge et al., 1999; Hanagasi et al., 2002; Hammer et al., 2011), resulting in impaired performance in tasks that rely on executive function.

Our results are also consistent with previous studies (Abrahams et al., 2000, 2004) that reported reduced verbal fluency that relies on executive dysfunction in ALS patients. The authors found that language dysfunction also plays a role in word retrieval deficits in ALS patients, reflected by hypoactivity in the middle and superior temporal gyri, which are involved in lexical and semantic processing (Abrahams et al., 2004). Our findings also suggested impaired lexical processing in ALS patients, reflected by decreased activation of the left superior and middle temporal gyri in the P2 time window, which appeared to be independent of executive dysfunction, reflected by decreased activity in the ACC and medial frontal gyrus in the P3 and N4 time windows.

Our findings appear to be partially inconsistent with Goldstein et al. (2011), who reported increased activity in these areas during the performance of the classic Stroop task but a decrease during a modified version of the same task. However, these two studies are only partially comparable. In fact, in the study by Goldstein et al. the subjects had to name the color of the ink in which the stimuli were presented. To provide the correct response, the subjects had to inhibit the automatic reading of the incongruent color word. Thus, the results reported account for altered inhibitory processing in ALS.

In the present study, mental discrimination was used to avoid movement or speech impairment contamination of the results. The subjects simply had to detect stimulus conflict that arose only at the level of semantic encoding, which contributes to overall interference in the classic Stroop task, together with the response level conflict (Milham et al., 2001; van Veen and Carter, 2005). The subjects also had to discriminate between congruent and incongruent stimuli without generating any response. Thus, our results more likely suggest impaired conflict-monitoring function in ALS patients.

In PLS patients compared with controls, the voxel-wise group comparison of ERPs using LORETA did not indicate differences in activity, in contrast to the findings in ALS patients. This result is consistent with the ERP and RT results.

Further activation studies that directly compare ALS and PLS patients are needed to better clarify differences in cognitive function between these two MNDs.

Previous studies compared cerebral involvement in ALS and PLS patients (Ciccarelli et al., 2009) and found a reduction of FA in ALS patients in white matter adjacent to the superior frontal gyrus. In PLS patients, they found lower FA in the body of the corpus callosum and white matter adjacent to the right primary motor cortex. These results, although not directly comparable with our results, suggest the possibility that ALS and PLS present a different pattern of cerebral involvement, which is consistent with our present findings.