Content-based clustering of hallucinations across sensory modalities in a large online survey

Abstract

Abstract Hallucinations can have rather heterogeneous aetiology and presentation. This inspired the concept of different subtypes based on symptom profiles, especially in the field of auditory hallucinations. As many people experience hallucinations in more than one sensory modality, it seems important to investigate potential hallucination subtypes across different sensory modalities. We assessed the content of hallucinations as part of a large survey among the general Dutch population (n = 10,448) using the Questionnaire for Psychotic Experiences. Based on their descriptions, thematic categories were created in a data-driven cluster analysis. 2594 participants who experienced hallucinations over the past week that contained at least 2 different thematic categories were selected. Clustering of their hallucination content was performed with the HDBSCAN method. We identified 4 clusters, i.e., subtypes, which can be typified as 1. hallucinations of foul odors, 2. complex visual scenes, 3. a vast variety of rather common hallucinations possibly related to heightened alertness, and 4. possibly bereavement hallucinations. The bereavement subtype showed an increase in emotional loneliness and the presence of delusions. Our findings suggest that the content of hallucinations can be informative, especially when investigated across sensory modalities. Such subtypes may help to better understand their underlying mechanisms.

Keywords

• Hallucinations
• Multimodal hallucinations
• Multisensory hallucinations
• Clustering
• Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN)
• Questionnaire for Psychotic Experiences (QPE)