Barzykowski and Moulin (B&M) presented an interesting model of autobiographical memory integrating both involuntary autobiographical memories (IAMs) and déjà vu phenomena. We suggest analysing déjà vu phenomenon in a broader context of misidentification syndromes (MIS) and propose the inclusion of anterior thalamic nuclei (ANT), particularly relevant for source memory, and retrosplenial cortex (RSC) into the neuroanatomical basis of the model.

Firstly, the model proposed by B&M suggests the sequential nature of memory retrieval process as depicted in Figure 1. Drawing upon a clinical example of delusional MIS, we suggest that these processes are not necessarily sequential. While in Capgras syndrome (CS) access to memory content is preserved (patient's proxy face is correctly recognized), feeling of familiarity is missing (the proxy is perceived as an imposter). In contrast, patients with Fregoli syndrome misrecognize strangers as proxies, presenting with false feeling of familiarity. A recent case of highly selective CS supported the idea that right temporal lesions lead to dissociation of a familiar face with its emotional value (Nuara et al., Reference Nuara, Nicolini, D'Orio, Cardinale, Rizzolatti, Avanzini and De Marco2020), which is in line with double dissociation between overt face recognition (fusiform face area) and autonomic recognition (limbic cortex) hypothesized by Ellis and Lewis (Reference Ellis and Lewis2001). Of note, Darby, Laganiere, Pascual-Leone, Prasad, and Fox (Reference Darby, Laganiere, Pascual-Leone, Prasad and Fox2017) recently linked familiarity to left RSC and the belief evaluation to the right frontal cortex. The nearest posterior cingulate cortex was already linked to specific familiarity processing (Qin et al., Reference Qin, Liu, Shi, Wang, Duncan, Gong and Northoff2012) and autobiographical memory (Spreng, Mar, & Kim, Reference Spreng, Mar and Kim2009). Thus, as RSC is active during the retrieval of all types of autobiographical information – both emotionally laden and neutral (Vann, Aggleton, & Maguire, Reference Vann, Aggleton and Maguire2009) and linked to the limbic system (Powell et al., Reference Powell, Hindley, Nelson, Davies, Amin, Aggleton and Vann2018) we suggest its incorporation to the neuroanatomical basis of B&M's model.

As Capgras and Fregoli syndromes (or their analogues for place memory, e.g., reduplicative paramnesia) may be regarded as counterparts of jamais vu and déjà vu phenomena, it is surprising that the proposed model does not address the emotional component of familiarity with reference to relevant neural networks and that the other side of the coin – jamais vu phenomenon – is not thoroughly discussed.

We strongly believe that feeling of familiarity has its emotional aspect and cannot be reduced to (meta)-cognitive components. A broad meta-analysis of functional magnetic resonance imaging or positron emission tomography data involving a familiarity task by Horn et al. (Reference Horn, Jardri, D'Hondt, Vaiva, Thomas and Pins2016) evidenced the dependence of the involvement of emotional networks on the type of paradigm used. The first two, fully laboratory-based, paradigms are based on the initial encoding phase of stimuli that are not specifically familiar to the participant, and they activate only the prefrontal and parietal regions. However, the third paradigm, based on stimuli previously known to the participant is assumed to refer to self-related emotions and personal experience, activates the aforementioned regions and also parts of the limbic system (Horn et al., Reference Horn, Jardri, D'Hondt, Vaiva, Thomas and Pins2016). That revealed the undoubted role of emotional processing of stimuli of specific familiarity, and the paradigm used seems closer to the complexity of everyday functioning and potentially directly covers a larger group of clinical cases, such as the aforementioned delusional MIS.

The reduction of autobiographical memory model to its (meta)-cognitive components and the lack of inclusion of emotional networks may affect its explanatory value in the clinic, for example, in the case of MIS. Not only does a conviction than a known person or place has been replaced by its copy have emotional consequences for the affected person, but also this conviction itself is based on the impairment of emotional processing. It seems to us that déjà vu is also likely to be emotionally loaded. Thus, we postulate that the autobiographical memory model would benefit from emphasizing the role of neural networks that integrate memory for facts with their emotional value. The interactions between medial temporal-lobe structures and prefrontal medial cortex could have been given more importance.

Furthermore, the suggested model focusing on the normative aspects of déjà vu fails to include a stage of belief evaluation, leading to true versus false feeling of familiarity. Implementing a module reflecting belief evaluation into it, as operationalized by Darby et al. (Reference Darby, Laganiere, Pascual-Leone, Prasad and Fox2017), could improve the model's explanatory value in the memory clinic, especially in the context of MIS.

Secondly, we suggest further refinement of the neuroanatomical basis of déjà vu. In Table 1, B&M referred to medial temporal-lobe circuitry and cingulate cortex/prefrontal cortex. We would like to pinpoint the important role of ANT for source memory (particularly relevant for déjà vu phenomenon), as demonstrated in research on Korsakoff syndrome. The most characteristic symptom of Korsakoff's is severe declarative memory impairment with the prominent loss of the order of events over time (achronogenesis) or its temporal context (Kopelman, Reference Kopelman2015). Temporal encoding deficit in Korsakoff syndrome can occur independently of typical frontal pathology (Dillingham, Milczarek, Perry, & Vann, Reference Dillingham, Milczarek, Perry and Vann2021). Patients reveal distinct difficulties with source memory – the less efficient encoding of contextual temporal and also spatial information. While mammillary bodies specialize in direct and integrated action with ANT on recollective-based recognition, other diencephalic nuclei do not show an exclusive specialization in recognition. Nevertheless, ANT are more closely related to recall memory and the medial dorsal thalamic nuclei are hypothesized to be related to recognition memory (Aggleton, Dumont, & Warburton, Reference Aggleton, Dumont and Warburton2011). The crucial role of the ANT in amnesia linked to Korsakoff's syndrome, albeit hypothesized for years, was clearly evidenced over two decades ago (Harding et al., Reference Harding, Halliday, Caine and Kril2000). Recently, Segobin et al.'s (Reference Segobin, Laniepce, Ritz, Lannuzel, Boudehent, Cabé and Pitel2019) investigation using diffusion tensor imaging sequence analysis revealed that the disconnection between ANT and hippocampus, leading to ANT atrophy, was a neuroimaging marker of thalamic amnesia (Segobin et al., Reference Segobin, Laniepce, Ritz, Lannuzel, Boudehent, Cabé and Pitel2019).

Finally, we propose integrating the model by Barzykowski and Moulin with the recent Aggleton and O'Mara (Reference Aggleton and O'Mara2022), which assumes a parallel and partially convergent operation of two memory streams: Hippocampal–cortical and a medial diencephalic–cortical stream. Hippocampus remains the core element of the first stream, while ANT are at the heart of the second one. The interactions between these streams seem crucial for mnemonic consolidation (Aggleton & O'Mara, Reference Aggleton and O'Mara2022).