Abstract
Cognitive biases in the recognition of synthetic emotional speech in alexithymia. In this article, we research the cognitive biases in the recognition of synthetic emotional speech generated using formantbased text-to-speech conversion, related to emotion processing capacities in alexithymia, in order to assess that alexithymia reflects a deficit in the ability to process emotional stimuli. The different recordings of the stimuli were presented to the each listener (162 undergraduate students), and he/she had to identify the emotion. It was possible to choose among neutral, happy, sad, angry and unidentified. Operating on these results, we have obtained the confusion matrices of the identified emotions. Results show that the subjects categorized as alexithymics recognize significantly less emotional expressions that the subjects with low levels of alexithymia in the first half of the experiment. These findings suggest that alexithymics were less able to recognize emotional stimuli. Also these findings provide further support for the validity of the alexithymia construct.
