Testing the bottleneck account for post-error slowing beyond the post-error response.

The bottleneck account for post-error slowing assumes that cognitive resources are depleted after errors and thus the processing of subsequent events is delayed. To test this, we used a novel speeded-choice task and recorded behavioral measures and ERP (event-related potential) components on five trials following either an erroneous or correct response. We found that participants were slower and less accurate immediately after making an error and that this reduction of performance decayed on the following trials. Moreover, post-correct versus post-error differences in both the visual N1 and the P3 component were found. However, the difference in the P3 component rapidly diminished over time, whereas the differences in the N1 component were still evident in the fourth trial following the erroneous response. The results lay further support to the bottleneck account for post-error slowing and show a combination of early attentional and higher-order processing changes that occur after erroneous responses.

Response time distribution parameters show posterror behavioral adjustment in mental arithmetic.

After making an error, we usually slow down before our next response. This phenomenon is known as the posterror slowing (PES) effect. It has been interpreted to be an indicator of posterror behavioral adjustments and, therefore, has been linked to cognitive control. However, contradictory findings regarding PES and posterror accuracy cast doubt on such a relation. To determine whether behavior is adjusted after making an error, we investigated other features of behavior, such as the distribution of response times (RT) in a mental arithmetic task. Participants performed an arithmetic task with (Experiments 1 and 2) and without (Experiment 1) an accuracy-tracking procedure. On both tasks, participants responded more slowly and less accurately after errors. However, the RT distribution was more symmetrical on posterror trials compared to postcorrect trials, suggesting that a change in processing mode occurred after making an error, thus linking cognitive control to error monitoring, even in cases when accuracy decreased after errors. These findings expand our understanding on how posterror behavior is adjusted in mental arithmetic, and we propose that the measures of the RT distribution can be further used in other domains of error-monitoring research.

Response trajectories capture the continuous dynamics of the size congruity effect.

In a comparison task involving numbers, the size congruity effect refers to the general finding that responses are usually faster when there is a match between numerical size and physical size (e.g., 2-8) than when there is a mismatch (e.g., 2-8). In the present study, we used computer mouse tracking to test two competing models of the size congruity effect: an early interaction model, where interference occurs at an early representational stage, and a late interaction model, where interference occurs as dynamic competition between response options. In three experiments, we found that the curvature of responses for incongruent trials was greater than for congruent trials. In Experiment 2 we showed that this curvature effect was reliably modulated by the numerical distance between the two stimulus numbers, with large distance pairs exhibiting a larger curvature effect than small distance pairs. In Experiment 3 we demonstrated that the congruity effects persist into response execution. These findings indicate that incongruities between numerical and physical sizes are carried throughout the response process and result from competition between parallel and partially active response options, lending further support to a late interaction model of the size congruity effect.

The Neural Signatures of Processing Semantic End Values in Automatic Number Comparisons.

The brain activity associated with processing numerical end values has received limited research attention. The present study explored the neural correlates associated with processing semantic end values under conditions of automatic number processing. Event-related potentials (ERPs) were recorded while participants performed the numerical Stroop task, in which they were asked to compare the physical size of pairs of numbers, while ignoring their numerical values. The smallest end value in the set, which is a task irrelevant factor, was manipulated between participant groups. We focused on the processing of the lower end values of 0 and 1 because these numbers were found to be automatically tagged as the "smallest." Behavioral results showed that the size congruity effect was modulated by the presence of the smallest end value in the pair. ERP data revealed a spatially extended centro-parieto-occipital P3 that was enhanced for congruent versus incongruent trials. Importantly, over centro-parietal sites, the P3 congruity effect (congruent minus incongruent) was larger for pairs containing the smallest end value than for pairs containing non-smallest values. These differences in the congruency effect were localized to the precuneus. The presence of an end value within the pair also modulated P3 latency. Our results provide the first neural evidence for the encoding of numerical end values. They further demonstrate that the use of end values as anchors is a primary aspect of processing symbolic numerical information.

The cost of errors: Perceived error detection in dual-task conditions.

Detecting that an error has been made can be crucial for the implementation of appropriate behavioral adjustments. Brain imaging studies indicate that error detection is not limited to response errors and that similar mechanisms are engaged even when behavioral control is not needed. The current study examines whether perceived error detection - the detection of erroneous stimuli that violate our expectations - requires central resources. In two experiments - using a dual-task design - we show that perceived error detection in the first task creates a bottleneck in information processing and delays the response selection of the second task. The results suggest that the requirement for central cognitive resources is a general feature of error detection because it is present even when the demand for behavioral control is low.

Processing ordinality and quantity: ERP evidence of separate mechanisms.

We report an event-related potential (ERP) experiment of ordinal processing exploring the relationship between ordinal and numerical information. ERPs were recorded from healthy adults while making ordered/non-ordered judgments on 3 non-symbolic numerical stimuli (arrays of dots). Three main variables were manipulated: (1) Ordinality (ordered vs. non-ordered groups of dots), tapping the quick "gist" estimation of ordinality. (2) Direction (ascending vs. descending order), tapping the symbolic, culturally influenced aspect of ordinality, and (3) Ratio between the group of dots, tapping the processing of the basic numerosity information. Behavioral results showed independent effects for each variable, replicating our previous findings with this paradigm. ERP effects differentiated between three cognitive processes for estimating ordinality, processing numerosity, and direction. This differentiation was found both in terms of timing and topography: Order estimation was associated with early scalp parietal and lateral occipital positivity (80-130ms) originating in the left Middle Temporal Gyrus; numerical ratio was associated with a later scalp medial posterior positivity (130-200ms); and direction was associated with a late and widespread scalp right frontal and scalp right parietotemporal positivity and a corresponding scalp left frontal and scalp left parietotemporal negativity (300-600ms). A theoretical model is suggested, stressing an early and basic ordinal-specific mechanism.