For people with post-traumatic stress disorder (PTSD), recalling traumatic memories often displays as intrusions that differ profoundly from processing of ‘regular’ negative memories. [2/20] Yes, that's a DALL-E

This begs the question of whether traumatic memories are an exceptionally strong manifestation of autobiographical memory or a divergent neural representation of memory? [3/20]

Although personal autobiographical memory is at the core of PTSD, research has largely focused on non-personal paradigms. This incongruency is understandable given that autobiographical memories vary across people, making their incorporation into paradigms challenging. [4/20]

We designed a study that examines PTSD patients’ responses to their own personal traumatic memory in the form of a structured, annotated, audio narrative. We compared traumatic memory, within patients, to a negatively-valenced non-traumatic sad memory, and a calm memory. [5/20]

We hypothesized that across PTSD patients, semantic similarity would correspond to neural similarity: if the memories of two patients are semantically close, their patterns of neural responses while listening to the audio script of these memories should be similar too. [6/20]

However! if traumatic autobiographical memories diverge from — rather than being a version of — sad autobiographical memories, then we would observe the semantic-to-neural relationship only for sad, but not traumatic memories. [7/20]

We quantified similarity between narratives across patients with word embedding and observed that both types of negatively-valenced narratives (PTSD and sad) formed overlapping clusters in semantic space, whereas ‘calm’ ones clustered in a separate part of semantic space. [8/20]

To account for the idiosyncratic nature of autobiographical memories, we used intersubject representational similarity analysis (#ISRSA) to relate personalized semantic content of the scripts with neural representations of the hippocampus acquired while listening. [9/20]

This is a twisted use for ISRSA where both the neural responses and stimuli differ across participants! Yep. These are real autobiographical memories – no two are alike. Luckily semantic similarity is used to anchor them in a relational structure. [10/20]

Our main result is that hippocampal patterns showed a differentiation in semantic representation by narrative type; Sad (negative yet non-traumatizing) scripts which were semantically similar (e.g. of death of a loved one) across patients, elicited similar neural patterns [11/20]

Conversely, thematically-similar traumatic autobiographical memories did not elicit similar representations. In this ISRSA figure we show cross-participant semantic-to-neural mapping of PTSD and sad autobiographical memories (red and blue). [12/20]

We also had quite a few control analyses as well as a classifier but you’ll have to read the preprint to learn more :) You also must be wondering how did the amygdala check out in our ISRSA - read the preprint! :) [13/20]

When dividing the cohort by symptom severity, we found that individual symptom severity modulated the representation of the traumatic narratives in the posterior cingulate cortex (PCC), another system for processing of autobiographical memory, imagery, mental time travel! [14/20]

Why would traumatic memories be represented differently than non-traumatic ones? We speculate that dysfunctions in peritraumatic encoding, collectively referred to as ‘memory fragmentation' may be the culprit of the poorer semantic representation of the traumatic memory. [15/20]

Yes, that is what DALL-E gives you when you ask to draw a painting of 'memory fragmentation'. Neat.

The weak semantic-to-neural mapping of traumatic narratives we observed resonates with reports of memory impairment attributed to the traumatic experience – memory disorganization, difficulty in voluntary retrieval and lack of narrative coherence. [16/20]

Why is it cool? In clinical settings, evaluation of traumatic memory organization is often based on meta-memory: self-report of memory coherence for the experience. Our approach using IS-RSA to study personal memories may allow a more objective neural marker for PTSD. [17/20]

The very nature of PTSD phenomenology is an open question: is PTSD an extreme case of ‘standard’ negative emotional processing or a divergent cognitive entity? Our findings support the idea of a separate cognitive experience in the reactivation of traumatic memories. [18/20]

This has been a tremendous collaboration! Shout outs to our coauthors - @orduek @krkulkarni_ Ben Kelmendi, Shelley Amen, Charles Gordon, John Krystal, @ifat_levy @harpaz_l and Daniela Schiller! Thank you so much for allowing us to explore such a valuable dataset! [19/20]

Finally, thank you friends and colleagues for your help and insight on this project! @YeshurunYaara @aya_benya @AharonRavia and Tem Orederu [20/20]

@OferPerl Congrats Ofer! well done for your perseverance throughout these difficult COVID years 💪