2.3. Deictic distinctions task

During scanning, participants performed a listening task with auditory linguistic stimuli. Two conditions, low- and high-deixis, were alternated in a block-design fashion. In the ‘low-deixis’ condition, participants heard speech excerpts in the form of short narratives that were impersonal and fact-stating and exclusively used the grammatical third person in any of the noun phrases used. Narrative excerpts were adapted from alternative versions of the prose recall subtest of the Rivermead Behavioral Memory Test (RBMT, Kurtz 2018) and the logical memory subtest of the Wechsler Memory Scale (WMS-III, Wechsler 1997). The original stimuli and their translation can be found in the Supplementary Information. To make this low-deixis condition as similar as possible to the high-deixis condition, and given the phenomenology of AVH in general, two female speakers took conversational turns in describing the facts or events in question, yet they did not interact with or addressed each other. In the ‘high-deixis’ condition, stimuli consisted of short dialogues between the same two speakers covering the same mundane topics and vocabulary, but the speakers now interacted linguistically with each other, using the grammatical first and second Persons to refer to themselves and address their interlocutor. Moreover, they occasionally spoke directly to the listener (once per block) as well, with phrases such as ‘And you, who are listening, what do you think?’, thus covering the full three-fold deictic frame of a conversation. Stimuli across the two conditions did not differ in syntactic complexity viewed through the proxy of the rate of embedded clauses (Mann-Whitney U test = -1.938, p = 0.427).

Six blocks of each condition were presented in alternating order, lasting 28 s each and separated by 14 s baseline periods in which only white noise was presented. Stimuli were delivered through MRI-compatible headphones (VisuaStim Digital, Resonance Technology, Northridge, CA, USA). To maintain visual stimulation constant and similar for all participants, the task was performed with eyes open while looking at a gray blank screen shown through MRI-compatible goggles (VisuaStim Digital).

Participants were instructed to remain silent and listen carefully to the recordings during the task. To ensure they paid attention during the task, a brief questionnaire was administered immediately after the task about the topics covered in the dialogues and narratives. Participants also self-reported their level of attention and, in the case of AVH + patients, the frequency of hallucinations during the task. Participants who reported not paying attention or were unable to identify the topics covered in the task (as defined by making 2 + mistakes in response to the 6 yes/no questions) were excluded from the analyses.

Immediately before the scanning session, participants were given task instructions and were shown a practice example of a high-deixis and a low-deixis excerpt, which was not subsequently used in the scanner.