Distributed Representations Meeting, 15 March 2012 Irina Simanova
Materials and methods 7 epileptic patients, 16 sessions Sites: hyppocampus, amygdala, enthorinal cortex, parahippocampal cotrex Microwire electrodes: 8 recording channels and the reference objects (M = 16, SD = 3.1) presented, for each object: 3 pictures, a spoken name, a written name; 6 times each Firing units detected; also addressed if the firing was preferred to pictures (visual invariance) Out of 750 units 79 had a significant response to at least one stimulus
A neuron in the left anterior hippocampus
A neuron in the enthorinal cortex
A neuron in the amygdala
Response invariance Visual invariance: ROC analysis, p<0.01 Triple invariance: visual invariance + significant responses to spoken or visual names N responses to the individual / total number of pictures of this individual (3). N responses to the other pictures/ total number of other pictures. Amplitude thresholds:
Visual invariance
Triple invariance
Discussions:
Modality-independent decoding of conceptual information (fMRI study) (Simanova et al., submitted)
Searchlight SVM, generalization across the four modalities, group-level statistical results
Multimodal integration = conceptual representation ?