Self-concept is our mental representation of who we are (identity) and what we are like as a person (personality). This representation plays a central role in our life – it allows us to learn from past personal experiences, predict future encounters, and generate complex behaviors. However, the neuronal and computational mechanisms of how we represent ourselves remain largely unknown.
Apart from its general relevance to all of us as thinking individuals, self-concept is important from a clinical perspective. For example, depression is characterized by increased self-focus and negative self-evaluations, whereas people with autism have problems inferring about their own as well as other people’s mental states. Other maladies of the self include depersonalization, dissociative identity, and borderline personality. Current treatments of these disorders are not always effective, which poses serious personal and socioeconomic challenges. Formulating a neuroscientific model of the sense of self is a necessary step toward designing new, brain-stimulation-based therapies.
The aim of this EU project funded by the Marie Skłodowska-Curie Action is to use the state-of-the-art method of intracranial electrophysiology to identify and characterize single neurons in the human brain that represent conceptual information about the self. We record electrical brain activity from awake human subjects implanted with depth electrodes for pre-surgical assessment. Our cognitive experiments are conducted during this clinical monitoring period, with no added risk to the patients. This project is a collaboration between the Karolinska Institute (Sweden) and the University of California, Los Angeles (USA).
(A) Electrodes used for intracranial recordings. The intracranial EEG (iEEG) contacts are used to localize the epileptic activity, whereas local field potentials (LFPs) and extracellular spiking activity are recorded from the microwires protruding from the electrode tip. The spikes from neurons that are close to the electrode tip can be separated via spike sorting algorithms (see further); neurons further away from the tip can be detected but not sorted, thus generating the multiunit activity; neurons that are even more distant contribute to the background noise. (B) Computed tomography (CT) and (C) CT fused with MRI showing one of the electrodes implanted in the medial temporal lobe. Adapted from Quian Quiroga, Cell, 2019.
Main steps of spike-sorting algorithms. Adapted from Quian Quiroga, Current Biology, 2012.
Some neurons in the medial temporal lobe respond selectively and invariably to images of a specific person or object. For example, images of ‘‘Mr. T,’’ including his written name (highlighted), elicited responses in a neuron in the amygdala in a subject who was fan of the film Rocky III. The second and third rows correspond to the raster plot and the peristimulus time histogram, respectively. Dashed vertical lines show the time of picture onset and offset, 1 s apart. Adapted from Quian Quiroga, Cell, 2019.