Journal of Neuroscience Methods 203 (2012) 233–240 Contents lists available at SciVerse ScienceDirect Journal of Neuroscience Methods jou rnal h om epa ge: www.elsevier.com/locate/jneumeth A new (semantic) reflexive brain–computer interface: In search for a suitable classifier A. Furdea a,b,∗ , C.A. Ruf a , S. Halder a,b , D. De Massari a,c,d , M. Bogdan b,e , W. Rosenstiel b , T. Matuz a , N. Birbaumer a,d a Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, Germany b Wilhelm-Schickard Institute for Computer Engineering, University of Tübingen, Germany c Graduate Training Centre of Neuroscience, International Max Planck Research School, Tübingen, Germany d IRCCS, Ospedale San Camilo, Venezia-Lido, Italy e Computer Engineering, University of Leipzig, Germany a r t i c l e i n f o Article history: Received 7 July 2011 Received in revised form 14 September 2011 Accepted 15 September 2011 Keywords: Classical semantic conditioning Electroencephalogram Single-trial classification Brain–computer interface a b s t r a c t The goal of the current study is to find a suitable classifier for electroencephalogram (EEG) data derived from a new learning paradigm which aims at communication in paralysis. A reflexive semantic classi- cal (Pavlovian) conditioning paradigm is explored as an alternative to the operant learning paradigms, currently used in most brain–computer interfaces (BCIs). Comparable with a lie-detection experiment, subjects are presented with true and false statements. The EEG activity following true and false statements was classified with the aim to separate covert ‘yes’ from covert ‘no’ responses. Four classification algorithms are compared for classifying off-line data collected from a group of 14 healthy participants: (i) stepwise linear discriminant analysis (SWLDA), (ii) shrinkage linear discriminant analysis (SLDA), (iii) linear support vector machine (LIN-SVM) and (iv) radial basis function kernel support vector machine (RBF-SVM). The results indicate that all classifiers perform at chance level when separating conditioned ‘yes’ from conditioned ‘no’ responses. However, single conditioned reactions could be successfully classified on a single-trial basis (single conditioned reaction against a baseline interval). All of the four investigated clas- sification methods achieve comparable performance, however results with RBF-SVM show the highest single-trial classification accuracy of 68.8%. The results suggest that the proposed paradigm may allow affirmative and negative (disapproving negative) communication in a BCI experiment. © 2011 Elsevier B.V. All rights reserved. 1. Introduction Brain–computer interfaces (BCIs) are devices that allow users to convey messages or commands without using the brain’s motor output pathways. Thus, BCIs provide a non-muscular communica- tion channel for individuals who are no longer able to communicate due to severe physical impairment. Neurological diseases such as amyotrophic lateral sclerosis (ALS), muscular dystrophy, locked-in syndrome (LIS) or high spinal cord injury may lead to severe or complete motor paralysis making communication difficult or even impossible. In locked-in state (LIS) severely paralyzed patients have resid- ual voluntary control over particular muscles (e.g. eye muscles, lips, fingers) (Birbaumer, 2006b). Patients may, however, develop the completely locked-in state (CLIS) in which all motor control is lost ∗ Corresponding author. Tel.: +49 7071 29 78295; fax: +49 7071 29 5956. (Birbaumer, 2006b). Over the last years it has been shown that patients with severe motor disability and also patients in the LIS are able to control an electroencephalography (EEG) based BCI (e.g. to select letters and thus to communicate) by regulating slow corti- cal potentials (SCP), sensori-motor rhythm (SMR) or with the P300 event-related potential (ERP) (Birbaumer et al., 1999; Kübler et al., 2005; Sellers et al., 2006; Neuper et al., 2003; Nijboer et al., 2008). There are no documented cases of CLIS patients communi- cating by means of BCI. In their meta-analysis of 29 patients in different stages of physical impairment who were trained with BCIs, Kübler and Birbaumer showed that none of the seven CLIS patients ever achieved any BCI control despite intact passive cogni- tive functioning, assessed with a battery of cognitive event-related potential-tests (Kübler and Birbaumer, 2008; Kotchoubey et al., 2003a,b). Therefore, one of the most challenging goals of BCI research remains the restoration of communication in CLIS. It has been suggested that a paradigm shift from instrumental-operant learning to classical conditioning is necessary to overcome the fail- ure of CLIS patients to achieve BCI control (Birbaumer, 2006a). 0165-0270/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jneumeth.2011.09.013