Flanker-Task in Children Time-Frequency Analyses of Response Monitoring Björn Albrecht 1,2 , Hartmut Heinrich 3,4 , Daniel Brandeis 2,5,6 , Henrik Uebel 1 , Juliana Yordanova 7 , Vasil Kolev 7 , Aribert Rothenberger 1 , and Tobias Banaschewski 1,2 1 Child and Adolescent Psychiatry, University of Göttingen, Germany, 2 Department of Child and Adolescent Psychiatry and Psychotherapy, Central Institute of Mental Health Mannheim, Germany, 3 Child and Adolescent Psychiatry, University of Erlangen, Germany, 4 Heckscher-Klinik, München, Germany, 5 Department of Child and Adolescent Psychiatry, University of Zurich, Switzerland, 6 Center for Integrative Human Physiology, University of Zurich, Switzerland, 7 Institute of Neurobiology, Bulgarian Academy of Sciences, Sofia, Bulgaria Abstract. Response processing may comprise multiple systems working in parallel at different functional levels of performance moni- toring. In time-frequency decompositions of response-locked event-related potentials from adults, a subprocess operating in the delta frequency band was interpreted as an index of cognitive error monitoring, distinguishable from a process with theta frequency probably related to motor control. However, it remains unclear whether such subprocesses can also be distinguished in children. In the current study, error processing was assessed in 22 normal boys aged 8 to 15 years using an Erikson Flanker task. Performance data revealed the expected indices of conflicting task demands, such as increased reaction times and error rates. A clear error-negativity was found in the response-locked event-related potentials after incompatible stimuli, and correct responses show a slow negative deflection immediately preceding the button-press, which is absent in errors. Time-frequency decompositions disclosed that a subprocess in the lower delta band preceding correct responses may reflect a more general action monitoring process sensitive to conflicting task demands that, moreover, may prevent one from making an error if it is active early enough. Processes in the delta and theta bands are modulated specifically by errors and may index motor-related monitoring in children. Moreover, these processes occurred considerably earlier for correct responses compared to errors, suggesting that their timing reflects some performance capacity. These considerations may help to clarify response processing in tasks with conflicting demands. Keywords: children, error processing, response monitoring, Ne, ERN, oscillations, wavelet, Flanker task Introduction Conflicting task demands and errors require action moni- toring in order to enable performance adjustments. Some 40 years ago, Rabbitt reported that errors in simple choice reaction time tasks were faster than correct responses, but that responses following errors were considerably slower. Thus, errors may occur when “a subject attempts to respond faster than some limitation to his capacity allows” (p. 272), which may further be followed by some kind of remedial adaptation (Rabbitt, 1966). Beginning in the 1990s, error processing was assessed using event-related potentials (ERPs, Falkenstein, Hohnsbein, Hoormann, & Blanke, 1990; Gehring, Goss, Coles, Meyer, & Donchin, 1993). To- day, there is consensus that performance errors are gener- ally accompanied by a negative ERP (error negativity or error-related negativity, N e ) peaking at approximately 100 ms after the erroneous response at fronto-central sites. N e is described in a variety of simple reaction tasks (Fal- kenstein, Hoormann, Christ, & Hohnsbein, 2000) and in tasks with more complex demands such as mental rotation (Band & Kok, 2000), when errors of choice or errors of commission (Scheffers, Coles, Bernstein, Gehring, & Don- chin, 1996) were made. It occurs when the response is given by hand, by foot (Holroyd, Dien, & Coles, 1998), or with eye movements (Nieuwenhuis, Ridderinkhof, Blom, Band, & Kok, 2001). Thus, several hypotheses ascribe N e a crucial role in error detection and action monitoring, such that it may reflect mismatch (Falkenstein et al., 1990; Gehring et al., 1993) or conflict (Carter et al., 1998) between error and required response. Dipole modeling further suggests there is a generator of N e in the anterior cingulate cortex and supplementary motor area (Dehaene, Posner, & Tucker, 1994; Herrmann, Rommler, Ehlis, Heidrich, & Fallgatter, 2004; Holroyd et al., 1998; Ullsperger & von Cramon, 2001; Van Veen& Carter, 2002). Recently, knowledge about N e has been extended by demonstrating that the classical time domain Ne signal is composed of multiple subcomponents reflecting distinctive aspects of behavioral monitoring (Yordanova, Falkenstein, Hohnsbein, & Kolev, 2004). In that study, a four choice reaction time task with either visual or acoustic stimuli was DOI 10.1027/0269-8803.23.4.183 © 2009 Federation of European Psychophysiology Societies Hogrefe Publishing Journal of Psychophysiology 2009; Vol. 23(4):183–190