Limits...
Internal and external information in error processing.

Heldmann M, Rüsseler J, Münte TF - BMC Neurosci (2008)

Bottom Line: When performance monitoring based on self-generated information was sufficient to detect a criterion violation an ERN was released, while the subsequent feedback became redundant and therefore failed to trigger an FRN.In contrast, an FRN was released if the feedback contained information which was not available before and action monitoring processes based on self-generated information failed to detect an error.The described pattern of results indicates a functional interrelationship of response and feedback related negativities in performance monitoring.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Neurology II, Otto-von-Guericke University Magdeburg, 39120 Magdeburg, Germany. marcus.heldmann@med.ovgu.de

ABSTRACT

Background: The use of self-generated and externally provided information in performance monitoring is reflected by the appearance of error-related and feedback-related negativities (ERN and FRN), respectively. Several authors proposed that ERN and FRN are supported by similar neural mechanisms residing in the anterior cingulate cortex (ACC) and the mesolimbic dopaminergic system. The present study is aimed to test the functional relationship between ERN and FRN. Using an Eriksen-Flanker task with a moving response deadline we tested 17 young healthy subjects. Subjects received feedback with respect to their response accuracy and response speed. To fulfill both requirements of the task, they had to press the correct button and had to respond in time to give a valid response.

Results: When performance monitoring based on self-generated information was sufficient to detect a criterion violation an ERN was released, while the subsequent feedback became redundant and therefore failed to trigger an FRN. In contrast, an FRN was released if the feedback contained information which was not available before and action monitoring processes based on self-generated information failed to detect an error.

Conclusion: The described pattern of results indicates a functional interrelationship of response and feedback related negativities in performance monitoring.

Show MeSH

Related in: MedlinePlus

ERPs time locked to correct responses for all timing conditions. Left panel: Bandpass filtered 4–12 Hz grand averages and isovoltage maps for correct responses of all time conditions (see text for details). Color scale represents steps of 0.1 microV. Right panel: Grand averages for identical conditions without bandpass filtering.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2291472&req=5

Figure 3: ERPs time locked to correct responses for all timing conditions. Left panel: Bandpass filtered 4–12 Hz grand averages and isovoltage maps for correct responses of all time conditions (see text for details). Color scale represents steps of 0.1 microV. Right panel: Grand averages for identical conditions without bandpass filtering.

Mentions: Trials with RTs faster than the response deadline were classified as „early“ (EA), trials with RTs slower than the response deadline as "early late" (EL) or "late late" (LL) trials. For each block this differentiation was made by computing the median of all RTs exceeding the individual response deadline. Thus, EL trials were trials slower than the deadline but faster than the median, while LL trials were those trials exceeding the median RT of the late responses. Additionally, for each time bin trials were categorized into correct and incorrect responses. Only the incorrect trials of the EA bin were used for analysis of the effects of choice errors, while in the remainder of the paper EA, EL and LL conditions will refer to responses with correct button selection only. After categorizing trials according to the above criteria response-locked (epoch length 900 ms, 300 ms baseline) and feedback-locked (epoch length 700 ms, 100 ms baseline) averages were calculated for each subject. To remove slow wave potentials like the P3 [30] data were subjected to a 4 – 12 Hz band pass filter (half amplitude cut-off, the effect of a 4–12 Hz filter can be seen in figure 3). Finally, filtered averages of all seventeen subjects were collapsed to calculate the grand averages. For all ANOVAs mean amplitudes (0–120 ms for response locked, 260–320 ms for feedback locked data) were used. In order to test the influence of incorrect button press responses on ERN and FRN-amplitude by itself an ANOVA with the factors correct/incorrect (2 levels) and electrode site (Fz, Cz, Pz) was performed for response- and feedback locked data of the early (EA) RT-bin. To assess the influence of the subjects' ability to detect reaction time errors of correct button press responses on ERN- and FRN-amplitudes, ANOVAs with the factors RT-bin (EA, EL, LL) and electrode site (Fz, Cz, Pz) were performed for the two components. To confirm the assumed interaction between ERN/FRN amplitude and availability of performance related information we calculated a repeated measures ANOVA with the factors electrode site (levels Fz, Cz), ERN/FRN (2 levels) and the factor RT-bin (3 levels). Prior to this analysis data were subjected to a vector-normalization procedure [31,32] to remove overall amplitude differences of FRN and ERN. The Huynh-Feldt epsilon coefficient was applied to correct ANOVAs for non-sphericity. The original degrees of freedom but corrected p-values will be reported.


Internal and external information in error processing.

Heldmann M, Rüsseler J, Münte TF - BMC Neurosci (2008)

ERPs time locked to correct responses for all timing conditions. Left panel: Bandpass filtered 4–12 Hz grand averages and isovoltage maps for correct responses of all time conditions (see text for details). Color scale represents steps of 0.1 microV. Right panel: Grand averages for identical conditions without bandpass filtering.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2291472&req=5

Figure 3: ERPs time locked to correct responses for all timing conditions. Left panel: Bandpass filtered 4–12 Hz grand averages and isovoltage maps for correct responses of all time conditions (see text for details). Color scale represents steps of 0.1 microV. Right panel: Grand averages for identical conditions without bandpass filtering.
Mentions: Trials with RTs faster than the response deadline were classified as „early“ (EA), trials with RTs slower than the response deadline as "early late" (EL) or "late late" (LL) trials. For each block this differentiation was made by computing the median of all RTs exceeding the individual response deadline. Thus, EL trials were trials slower than the deadline but faster than the median, while LL trials were those trials exceeding the median RT of the late responses. Additionally, for each time bin trials were categorized into correct and incorrect responses. Only the incorrect trials of the EA bin were used for analysis of the effects of choice errors, while in the remainder of the paper EA, EL and LL conditions will refer to responses with correct button selection only. After categorizing trials according to the above criteria response-locked (epoch length 900 ms, 300 ms baseline) and feedback-locked (epoch length 700 ms, 100 ms baseline) averages were calculated for each subject. To remove slow wave potentials like the P3 [30] data were subjected to a 4 – 12 Hz band pass filter (half amplitude cut-off, the effect of a 4–12 Hz filter can be seen in figure 3). Finally, filtered averages of all seventeen subjects were collapsed to calculate the grand averages. For all ANOVAs mean amplitudes (0–120 ms for response locked, 260–320 ms for feedback locked data) were used. In order to test the influence of incorrect button press responses on ERN and FRN-amplitude by itself an ANOVA with the factors correct/incorrect (2 levels) and electrode site (Fz, Cz, Pz) was performed for response- and feedback locked data of the early (EA) RT-bin. To assess the influence of the subjects' ability to detect reaction time errors of correct button press responses on ERN- and FRN-amplitudes, ANOVAs with the factors RT-bin (EA, EL, LL) and electrode site (Fz, Cz, Pz) were performed for the two components. To confirm the assumed interaction between ERN/FRN amplitude and availability of performance related information we calculated a repeated measures ANOVA with the factors electrode site (levels Fz, Cz), ERN/FRN (2 levels) and the factor RT-bin (3 levels). Prior to this analysis data were subjected to a vector-normalization procedure [31,32] to remove overall amplitude differences of FRN and ERN. The Huynh-Feldt epsilon coefficient was applied to correct ANOVAs for non-sphericity. The original degrees of freedom but corrected p-values will be reported.

Bottom Line: When performance monitoring based on self-generated information was sufficient to detect a criterion violation an ERN was released, while the subsequent feedback became redundant and therefore failed to trigger an FRN.In contrast, an FRN was released if the feedback contained information which was not available before and action monitoring processes based on self-generated information failed to detect an error.The described pattern of results indicates a functional interrelationship of response and feedback related negativities in performance monitoring.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Neurology II, Otto-von-Guericke University Magdeburg, 39120 Magdeburg, Germany. marcus.heldmann@med.ovgu.de

ABSTRACT

Background: The use of self-generated and externally provided information in performance monitoring is reflected by the appearance of error-related and feedback-related negativities (ERN and FRN), respectively. Several authors proposed that ERN and FRN are supported by similar neural mechanisms residing in the anterior cingulate cortex (ACC) and the mesolimbic dopaminergic system. The present study is aimed to test the functional relationship between ERN and FRN. Using an Eriksen-Flanker task with a moving response deadline we tested 17 young healthy subjects. Subjects received feedback with respect to their response accuracy and response speed. To fulfill both requirements of the task, they had to press the correct button and had to respond in time to give a valid response.

Results: When performance monitoring based on self-generated information was sufficient to detect a criterion violation an ERN was released, while the subsequent feedback became redundant and therefore failed to trigger an FRN. In contrast, an FRN was released if the feedback contained information which was not available before and action monitoring processes based on self-generated information failed to detect an error.

Conclusion: The described pattern of results indicates a functional interrelationship of response and feedback related negativities in performance monitoring.

Show MeSH
Related in: MedlinePlus