摘要：Brain-computer interfaces (BCIs) are a rapidly expanding feld of study and require accurate and reliable real-time decoding of patterns of neural activity. These protocols often exploit selective attention, a neural mechanism that prioritises the sensory processing of task-relevant stimulus features (feature- based attention) or task-relevant spatial locations (spatial attention) . Within the visual modality, attentional modulation of neural responses to diferent inputs is well indexed by steady-state visual evoked potentials (SSVEPs) . These signals are reliably present in single-trial electroencephalography (EEG) data, are largely resilient to common EEG artifacts, and allow separation of neural responses to numerous concurrently presented visual stimuli . To date, eforts to use single-trial SSVEPs to classify visual attention for BCI control have largely focused on spatial attention rather than feature- based attention . Here, we present a dataset that allows for the development and benchmarking of algorithms to classify feature-based attention using single-trial EEG data . The dataset includes EEG and behavioural responses from 30 healthy human participants who performed a feature-based motion discrimination task on frequency tagged visual stimuli .
