Cortical hubs of highly superior autobiographical memory

Autobiographical


Introduction
Autobiographical memory, an individual's recollection of personal experiences and events from their life, forms a crucial part of human cognition, linked to the sense of selfidentity (Conway & Pleydell-Pearce, 2000).The importance of autobiographical memory extends beyond identity formation; it also influences decision-making processes and future behavior based on past experiences (Schacter, 2012;Schacter et al., 2015).The neurobiological correlates associated with the retrieval of autobiographical memory reflect the complexity of this process, involving a distributed network of brain regions, including the anterior and posterior cortical midline regions, the medial temporal lobe (MTL), the temporal pole, and the lateral posterior parietal cortex (for reviews, see Boccia et al., 2019;Cabeza & St Jacques, 2007;Daviddi et al., 2023;Svoboda et al., 2006).While brain activation within these regions has been extensively investigated in people with typical autobiographical memory capacity, recent studies have revealed high individual variability in the capability to recollect personal experiences (Palombo et al., 2018), ranging from individuals with "severely deficient" (Conti et al., 2023;Palombo et al., 2015) to "highly superior" autobiographical memory (HSAM; Parker et al., 2006;LePort et al., 2012).The investigation of these "extreme" conditions can significantly contribute to our understanding of autobiographical memory, providing valuable insights into the neurobiological basis of human memory (e.g., Santangelo et al., 2022).Here, we leverage graph theory methods to describe alterations of resting-state brain connectivity in HSAM individuals.
HSAM is a rare condition characterized by the ability to vividly recall personal events from the past with specific details, such as the date, time, and location of the recalled event (Parker et al., 2006).This phenomenon has received considerable attention in the fields of cognitive psychology and cognitive neuroscience, with a growing number of studies describing the neural correlates associated with enhanced autobiographical memory (LePort et al., 2012;Mazzoni et al., 2019;Santangelo et al., 2018Santangelo et al., , 2020Santangelo et al., , 2021)).LePort et al. (2012) conducted a comprehensive investigation using behavioral and neuroanatomical measures, finding distinct characteristics in individuals with HSAM compared to control participants.They reported that HSAM individuals show different neuroanatomical correlates than controls in several brain regions encompassing -among others-the medial temporal and the posterior parietal cortex.A subsequent study examined the functional correlates of HSAM, finding enhanced brain activity associated with memory access in individuals with HSAM, suggesting that their superior memory ability may be related to more efficient retrieval processes (Santangelo et al., 2018).Additionally, research has identified a link between obsessive-compulsive disorder (OCD) and HSAM (LePort et al., 2012;Santangelo et al., 2018), a connection further underscored by findings of abnormal resting-state effective connectivity in OCD patients (Xu et al., 2024).It may be the case that similar functional connectivity patterns may underlie both conditions and contribute to the remarkable mnemonic abilities observed in HSAM individuals.Current research has thus provided some initial evidence of the neurobiological mechanisms associated with the enhanced retrieval of autobiographical memories in HSAM; however, brain functioning associated with HSAM when these individuals are not performing a task requiring autobiographical memory retrieval (i.e., during resting-state functional neuroimaging) remains poorly understood.Studies of resting-state fMRI data have identified altered functional connectivity in HSAM, though this work has largely focused on individual case studies (Ally et al., 2013;Brandt & Bakker;2018;De Marco et al., 2021;see Daviddi et al., 2022a for a group study).
Recent meta-analyses have emphasized the involvement of cortical midline regions, distributed within the default network (DN), in autobiographical and self-related processes (Araujo et al., 2013;Daviddi et al., 2023;Martinelli et al., 2013;Svoboda et al., 2006;Talbot et al., 2024).The DN is a large-scale brain network comprised of distributed brain regions, including the medial prefrontal cortex and posterior cingulate cortex, that are engaged when individuals remember past experiences, imagine future experiences, or engage in related forms of mental simulation (Buckner & DiNicola, 2019;Schacter et al., 2007).DN connectivity is thought to play a critical role in episodic memory, and damage to the DN has been shown to disrupt autobiographical memory retrieval (Philippi et al., 2015).Accordingly, we hypothesize that HSAM individuals will show enhanced connectivity within regions of the DN within posterior and anterior regions of the cortical midline, known to be involved in autobiographical memory.
Graph theory analyses have proven to be effective methods for exploring the functional networks of the human brain (van den Heuvel & Hulshoff Pol, 2010), and within this framework, various metrics reveal different aspects of the cerebral networks.We selected weighted degree (WD) connectivity as it represents a comprehensive measure of node centrality that combines both the presence and the strength of functional connections (Buckner et al., 2009;Diez & Sepulcre, 2018;Sepulcre, 2014;Sepulcre et al., 2010).WD is particularly apt for assessing the integral roles of the human cortex in HSAM, as it captures global communication and integration across the brain (Rubinov & Sporns, 2010).Moreover, WD connectivity offers robust test-retest reliability, providing a dependable framework for investigating the brain's complex network architecture (Xiang et al., 2020).Thus, in line with recent work, we compute WD to identify differences in brain network topology associated with autobiographical memory.
The present study leverages voxel-level WD connectivity to examine the network architecture related to an enhanced form of autobiographical memory.We describe connectivity differences between HSAM and controls using a dual approach.First, using a hypothesis-driven approach, we defined regions of interest within the autobiographical memory network based on independent studies with Neurosynth (Yarkoni et al., 2011) and examined seed-based connectivity from these regions of interest.Second, we sought to capture cortical hubs eareas of convergence that combine information from distributed regions across the braine defining the brain connectome of HSAM individuals without imposing any a priori region of interest.Through the implementation of seed-based and whole-brain approaches (Fig. 1A), we describe patterns of resting-state network topology associated with enhanced autobiographical memory.

Sample
The current study included a set of twelve individuals with HSAM (n ¼ 12; mean age ¼ 35.33; range 20e60) that agreed to participate.These individuals belonged to a larger sample of individuals with HSAM previously screened in the Italian population in accordance with the previous literature (LePort et al., 2012(LePort et al., , 2016(LePort et al., , 2017)).The screening procedure consisted of the Public Events Quiz and the Random Dates Quiz, both administered via telephone interviews, with no time limits.The Public Events Quiz consists of thirty questions based on public events selected from five categories: sporting events, political events, notable negative events, events involving famous people, and holidays.Fifteen of these questions ask the participant to retrieve the date of a given significant public event (national or international) (e.g., "Please state the day of the week and the exact date with the day, month, and year when Federica Pellegrini, the famous Italian swimmer, won the gold medal at the Olympic Games in Beijing"); the remaining fifteen questions ask the participant to associate a given date with a highly significant public event (e.g., "What happened on June 25, 2009?").For each question, the participant is asked to identify the day of the week on which the date occurred.All questions are about events that occurred when that participant was at least 8 years old.One point is awarded for each correct answer, and the maximum total score is 88 points.
The Random Dates quiz consist of eighteen computergenerated random dates ranging from the participant's age of fifteen to the day before the test.For each date, the participant is asked to provide three items: (1) the day of the week (2) a description of a verifiable event (i.e., any event that could be confirmed using a search engine); that occurred within one month before and after the generated date; (3) a description of a personal autobiographical event.One point is awarded for each correct day of the week, correct public event, and unverified personal autobiographical event.A maximum of three points could be earned per date (54 points total).The individuals with HSAM inluded in the study averaged 55% accuracy on the Public Event Quiz and 65% accuracy on the Random Dates Quiz, in line with the previous literature (e.g., LePort et al., 2012;Santangelo et al., 2021).The study also included a set of twenty-nine age-and sex-matched control participants (n ¼ 29, mean age ¼ 35.66; range 21e59) who were recruited through contacts in the community, none of whom reported having HSAM or other superior memory abilities (for a similar approach, see LePort et al., 2012).All participants gave written informed consent to the study, which was approved by the independent Ethics Committee of the IRCCS Santa Lucia Foundation (CE/PROG.540).We report how we determined our sample size, all data exclusions, all exclusion criteria, whether exclusion criteria were established prior to analysis, all manipulations, and all measures in the study.No part of the study procedures or study analysis was preregistered prior to the research being conducted.Analysis codes and scripts are available on the publicly accessible digital repository 'Open Science Framework' (OSF; https://osf.io/4eb7v/).

MRI acquisition & preprocessing
Scanning was conducted at the IRCCS Santa Lucia Foundation Neuroimaging Laboratory on a 3T Siemens Prisma scanner.Functional and anatomical magnetic resonance images were acquired using a quadrature volume head coil for radio frequency transmission and reception.Head movement was minimized by mild restraint and cushioning.High-resolution T1-weighted scans (MP-RAGE; TR/TE ¼ 2500/2, 240 Â 256 matrix, 1 mm thick, 176 mm FoV) were acquired for anatomical segmentation and transformation of functional images to standard space.Resting-state functional scans were acquired with a T2*-weighted echo-planar plus sequence with 32 interleaved slices, including 322 volumes [60 slices covering the whole brain, 2.42 Â 2.42 Â 2.4 mm, repetition time ¼ 1.1 sec, time echo ¼ 30 ms].MRI data for both anatomical and functional images were preprocessed using FMRIB Software Library v5.0.7 (FSL) and MATLAB 2017a (Mathworks Inc., Natick, MA).The anatomical and functional preprocessing pipelines were adapted from previous work (Diez et al., 2019).The anatomical T1 preprocessing included: reorientation to right posterior-inferior (RPI); alignment to anterior and posterior commissures; skull stripping; gray matter, white matter, and cerebrospinal fluid segmentation; and computation of non-linear transformation between individual skull-stripped T1 and 2 mm resolution MNI152 template images.The functional MRI preprocessing pipeline included: slice time correction; reorientation to RPI; realigning functional volumes within runs with a rigid body transformations (6 parameters linear transformation); computation of the transformation between individual skull-stripped T1 and mean functional images; intensity normalization; removal of confounding factors from the data using linear regression -including 12 motion-related covariates (rigid motion parameters and its derivatives), linear and quadratic terms, and five components each from the cerebrospinal fluid and white matter.Global signal regression was not applied due to the spurious correlations this can introduce (Murphy et al., 2009).Transformation of resting state data to MNI space was performed, concatenating the transformation from functional to structural and from structural to MNI, spatial smoothing with an isotropic Gaussian kernel of 6-mm FWHM, and band-pass filtering (.01e.08 Hz) to reduce low-frequency drift and high-frequency noise were also applied.Head motion was quantified using realignment parameters obtained during image preprocessing, including 3 translation and 3 rotation estimates.Scrubbing of time points with excess head motion was used to remove all time points with a frame displacement >.2 mm.To achieve connectivity estimates using the same number of time points between participants, the first 285 time points without movement were used.Two participants in the control group were removed from the study due to excessive head motion.The distributions of the correlations across time series were reviewed for possible contamination; no outliers were observed from connectivity distributions.

ROI definition
To initially characterize the specific network regions implicated in HSAM, we identified regions of interest that are commonly activated during autobiographical memory tasks.Seeds were derived using a voxel-wise, meta-analytic approach using the publicly available Neurosynth (Yarkoni et al., 2011).We conducted a search with the keyword 'autobiographical' to extract a probabilistic map of voxels associated with autobiographical memory based on findings from 143 studies.The resulting map was downloaded and subsequently dissected into five contiguous clusters to obtain five seed regions: the medial prefrontal cortex (mPFC), posterior cingulate cortex (PCC), medial temporal lobe (MTL), angular gyrus (AG), and temporal pole (TP) (Fig. 1B).For each participant, voxels contained in each region were then used to compute seed-based connectivity maps.The cluster size and peak coordinates are also reported (Table 1).These regions have been studied extensively in relation to autobiographical memory (see Boccia et al., 2019;Cabeza & St Jacques, 2007;Daviddi et al., 2023;Svoboda et al., 2006).Lateral seed regions were restricted to the right hemisphere for the present analysis, though results replicated with bilateral seed maps.

Weighted Degree connectivity
To quantify the relative prominence of each voxel within largescale brain networks, voxel-level WD values were computed.WD is a measure of centrality, calculated as the sum of the strengths of functional connections between each voxel and the rest of the brain (Bullmore & Sporns, 2009).Following similar procedures as in previous works (Bueicheku ´et al., 2020;Sepulcre et al., 2012), we used a mask of 42,448 voxels covering the entire brain.WD analyses were conducted at the individual level by calculating Pearson productemoment correlation coefficients for the time course of all brain voxels in a pairwise manner.A r-to-z Fisher transformation was applied to the resulting correlation matrix and negative values were removed due to their controversial interpretation (Qian et al., 2018).After obtaining a higheresolution correlation matrix for each participant, we summed all the r-values in a voxel-wise manner to generate individual WD maps (Fig. 1C).For each participant, we generate five seed-based WD maps, and one whole-brain WD map.

Statistical analysis
General linear models were used to compute the group differences between HSAM and control groups using wholebrain WD maps or seed-based WD maps as input images.
Participant age and sex were included as covariates in all analyses.Whole-brain correction for multiple comparisons was calculated using Monte Carlo simulation with 10,000 iterations to estimate the probability of false positive clusters with a two-tailed p-value <.05 (3dClustSim, https://afni.nimh.nih.gov/).Cortical surfaces were visualized using the Connectome Workbench platform.Surface images were displayed using a color scale based on t-scores.Heat maps for the corrected t-statistic range from 1.96 to 3; correspondingly, all voxels highlighted in this cortical projection are significant at the p < .05threshold.

Connectivity from the autobiographical memory network
We performed seed-based analysis to identify differences in WD connectivity in HSAM vs. control participants, from specific regions of interest associated with autobiogaphical memory.Cluster size and peak activations are additionally reported in Table 2. From the mPFC seed, we observed higher connectivity to the visual cortex and anterior cingulate cortex in HSAM participants relative to controls (Fig. 2A).From the PCC seed, we found higher connectivity in HSAM participants to the anterior cingulate cortex, along with voxels in the medial prefrontal cortex, with the parahippocampal cortex, and with the inferior parietal cortex.Additionally, we observed reduced connectivity from the PCC seed to the superior parietal cortex in HSAM participants, relative to controls (Fig. 2B).From the MTL seed, we observed higher WD connectivity to voxels within the posterior cingulate cortex and parahippocampal regions.Furthermore, we found reduced connectivity from the MTL seed to the supplementary motor area, anterior cingulate cortex, and anterior insula (Fig. 2C).We observed higher WD connectivity from the AG seed to the ventromedial prefrontal cortex in HSAM participants, relative to controls (Fig. 2D).Lastly, from the TP seed, we observed increased connectivity to the anterior cingulate and visual cortex in HSAM relative to control groups (Fig. 2E).

Cortical hubs associated with HSAM
We performed whole-brain WD analysis to identify grouplevel cortical hub differences between HSAM and control participants.HSAM individuals showed increased WD connectivity across wide swaths of the cortex, particularly along cortical midline regions.Specifically, the WD of voxels within the medial prefrontal, anterior cingulate, and retrosplenial cortices was higher in HSAM relative to controls (Fig. 3).To visualize this effect, we plot the WD scores for the voxel with peak activation (t ¼ 6.71, z ¼ 5.39, P < .001)within the medial prefrontal cortex.Additionally, we observed a cluster of voxels within the left dorsolateral prefrontal cortex with increased WD in the HSAM compared to control conditions.HSAM was also positively associated with WD of voxels within the language network, including left-lateralized middle temporal and angular gyrus.We did not observe any regions of increased WD connectivity in the control relative to HSAM conditions.In sum, these results indicate that HSAM is associated with higher WD connectivity, particularly within cortical midline areas overlapping with core hubs of the default network.

Discussion
The present study applied graph theory methods to investigate alterations in functional brain connectivity associated with HSAM, revealing a distinctive neural signature that may underlie this exceptional cognitive phenomenon.First, the seed-based approach revealed increased connectivity in HSAM individuals, as compared to controls, from all of the core regions of the autobiographical memory network: that is, from the mPFC, the MTL, and PPC along midline regions, as well as from the AG and the TP, along the lateral regions.Nevertheless, it is important to note that when the MTL was used as the seed region, we observed decreased WD connectivity to the supplementary motor area, the dorsolateral prefrontal cortex, the anterior cingulate cortex, and the insula.Second, the cortical-hub approach (i.e., the whole-brain WD analysis), identified widespread enhanced WD connectivity in the HSAM (compared to control) participants, particularly along cortical midline regions.Taken together, these results underscore the role of cortical midline structures in supporting autobiographical memory and illuminate the brain network organization that underlies HSAM.
By revealing differential patterns of connectivity, this study provides a basis for reconciling the contrasting findings which comprise the current literature on the "intrinsic" functional connectivity in HSAM subjects during resting-state.For example, in a recent case study, higher expression of large-scale and region-to-region connectivity was revealed in a single HSAM subject (De Marco et al., 2021), suggesting that a complex reorganization of memory networks in HSAM could be identified as the critical neural alteration of this extraordinary memory capability.On the contrary, a resting-state study where the anterior and posterior portion of the hippocampus were used as seed regions for functional connectivity analyses, revealed decreased functional connectivity in HSAM vs. control subjects (Daviddi et al., 2022b).Specifically, in this Fig. 2 e Seed-Based Connectivity.Seed-based WD maps were computed from five a priori defined regions of interest: medial prefrontal cortex (mPFC), posterior cingulate cortex (PCC), medial temporal lobe (MTL), angular gyrus (AG), and the temporal pole (TP).We compared WD maps for HSAM and control participants to identify differential patterns of connectivity, from these seed regions, associated with enhanced autobiographical memory.Overall, results show increased connectivity from seed regions to cortical midline areas (including anterior cingulate, medial temporal and visual cortex) in HSAM relative to control participants.study, HSAM individuals show reduced hippocampal connectivity with high-level brain regions belonging to the saliency network (i.e., anterior cingulate cortex and bilateral insulae) and ventral fronto-parietal network, compared to controls.However, in the same study, HSAM showed increased hippocampal connectivity with sensory regions (i.e., fusiform gyrus and inferior temporal cortex).Our current findings are consistent with both of these studies.In line with the results of De Marco et al. ( 2021), we observed a general increase in the functional connectivity within the autobiographical memory network, especially along the midline regions.Additionally, we observed reduced coupling between the MTL seed (including the hippocampal cortex) and key regions involved in saliency detection, such as the anterior cingulate cortex and the insula, thus replicating the findings previously reported by Daviddi et al. (2022b).This latter finding supports the idea that HSAM may be supported by altered (i.e., diminished) functional connectivity between the MTL and high-level control networks which could account for the tendency of individuals with HSAM to encode and consolidate episodic information regardless current salience (cf.Daviddi et al., 2022b).Moreover, the spread of enhanced connectivity along the midline regions observed within our cohort of HSAM individuals (see Fig. 2) resonates with previous findings that implicate cortical midline structures in the retrieval and processing of autobiographical memories (Araujo, Kaplan, & Damasio;2013;Daviddi et al., 2023;Maguire, 2001;Summerfield et al., 2009).The present findings extend this work by suggesting that these regions also show enhanced connectivity in HSAM individuals at rest.In our cohort of HSAM individuals, we observed increased WD connectivity across voxels in the anterior and posterior cortical midline relative to controls.This finding points to the intriguing possibility that the same networks implicated in typical autobiographical memory exhibit distinct connectivity patterns in HSAM individuals compared to controls.These alterations suggest a heightened degree of integration and communication among key regions of the autobiographical memory network.
Additionally, our results give further support to previous findings suggesting a close overlap between the autobiographical memory network and DN (Daviddi et al., 2023;Ino et al., 2011;Schacter et al., 2007Schacter et al., , 2012)), and suggest that the enhanced memory ability of HSAM individuals could be sustained by DN activity.The DN is commonly known for its activation during internally focused tasks, including daydreaming, future planning, and constructing personal narratives (Andrews-Hanna et al., 2010;Buckner et al., 2008).The present study employed WD connectivity as a measure of network centrality to pinpoint the integral hubs within this network that are associated with HSAM.Despite the preliminary nature of this work, we have reported initial evidence that cortical hubs may support the superior autobiographical memory performance observed in HSAM.More specifically, we observed increased whole-brain WD connectivity in core regions of the DN (i.e., the mPFC and PCC) in HSAM vs. control.These regions are critical components of the DN, and our findings suggest that they play a central role in the memory capabilities that characterize HSAM.Moreover, the key regions of the autobiographical network used as seed regions for the seed-based connectivity overlap with the DN (Andrews-Hanna et al., 2010), and showed a pattern of increased WD connectivity with distributed cortical areas belonging to the DN along the midline brain areas (see Table 2), thus providing additional support for a large-scale DN connectivity in HSAMs.However, it is worth emphasizing that our study focused on resting-state connectivity, where dynamic task-related changes were not captured.Future studies should build upon task-based findings Fig. 3 e Whole-Brain Connectivity.WD across the medial prefrontal, anterior cingulate, retrosplenial and visual cortices was higher in HSAM participants relative to controls.Additionally, WD of voxels within the language network, including leftlateralized middle temporal and angular gyrus were significantly higher in HSAM participants relative to controls.This effect is further illustrated by the violin plot comparing WD values in HSAM vs. controls for the peak voxel (MNI: ¡4, 52, ¡12) located within the medial prefrontal cortex.
to assess network reconfigurations during active autobiographical memory retrieval (Santangelo et al., 2018) and episodic future thinking in HSAM.
The enhanced integration within DN midline structures observed in our HSAM cohort, is consistent with some findings from previous investigations of this population (LePort et al., 2012;though see Talbot et al., 2024), suggesting that superior autobiographical memory performance may be linked with the structural and functional integrity of these DN regions.This suggestion is further corroborated by lesion studies indicating that disruptions to the DN, especially in areas such as the mPFC and hippocampus, lead to impairments in autobiographical memory retrieval capabilities (Philippi et al., 2015).Recent analyses of transient epileptic amnesia have revealed that disruptions to these networks can precipitate profound autobiographical memory loss (Baker et al., 2021;Ukai et al., 2021); however, the variability in memory impairment severity observed across different cases suggests that compensatory mechanisms may help to preserve certain aspects of memory function.This resilience implies that while specific hubs are integral to autobiographical memory, the distributed nature of the network allows for adaptation and compensation, which may explain the variance in autobiographical memory impairment.It is worth noting that DN activation extends beyond memory processes, playing a pivotal role in various cognitive tasks including prospection, spatial navigation, and theory of mind (Buckner & DiNicola, 2019;Li et al., 2014;Spreng et al., 2009), as well as other non-cognitive processes (e.g., Mehnert et al., 2023;Salone et al., 2016).Our findings contribute to a deeper understanding of the neural underpinnings that distinguish HSAM, highlighting the DN as an important neural substrate for enhanced memory capabilities and as a target for future research and interventions aimed at improving memory function.
Severely deficient autobiographical memory (SDAM) offers an intriguing contrast to HSAM, characterized by an inability to vividly recollect personal past experiences (Palombo et al., 2015).A recent investigation by Conti et al. (2023) identified neurobiological correlates of SDAM, including alterations in connectivity and activity within areas that are largely overlapping with those implicated in HSAM.Notably, while HSAM is associated with increased connectivity within regions of the autobiographical memory network, findings suggest that SDAM individuals have reduced cortical thickness in these regions, including the retrosplenial complex, the lateral occipital cortex, and the angular gyrus.Similarly, Palombo et al. (2015) reported reduced hippocampal volume and decreased activation in the anterior/posterior midline regions, namely, the medial prefrontal cortex and precuneus, during autobiographical memory retrieval.This juxtaposition highlights a spectrum of autobiographical memory capacity, with HSAM and SDAM representing extreme ends.Further research into these contrasting memory conditions may provide valuable insights into the neurocognitive basis of autobiographical memory, and underscore the role of large-scale brain networks in remembering one's personal past.
Our findings also resonate with studies investigating the neural basis of exceptional memory performance.In a landmark study, Maguire et al. (2003) identified structural differences in the brains of individuals capable of memorizing large amounts of information (i.e., memory athletes) compared with controls, particularly in the posterior hippocampus, which these individuals engage through mnemonic strategies.Comparatively, our HSAM participants, without such formal mnemonic training, exhibit enhanced connectivity from the medial temporal lobe to posterior cingulate cortex and parahippocampal regions.Despite these differences, both groups demonstrate altered connectivity in brain regions crucial for memory processes (see Dresler et al., 2017 for findings on memory athletes).This parallel indicates that whether through natural propensity, as in HSAM, or through deliberate practice, as seen in memory athletes, exceptional memory capability is underpinned by distinctive neurobiological adaptations.Future research could further investigate how training and innate abilities interact and potentially converge at the level of neural network organization to facilitate different kinds of superior memory performances.
The identification of enhanced connectivity within certain cortical hubs in HSAM invites a discussion on the directionality of this relationshipdwhether these hubs contribute to the development of HSAM or emerge as a result of it.Recent findings suggest that repeated retrieval of autobiographical memories can induce measurable neural changes, particularly in regions like the anterior hippocampus and the ventromedial prefrontal cortex (Bradley et al., 2022;Gurguryan et al., 2021).These studies propose a neural adaptation mechanism to repeated memory retrieval, raising the possibility that the pronounced connectivity within cortical hubs observed in HSAM individuals might also be a product of persistent and detailed autobiographical recollection.However, it remains unclear whether the distinctive connectivity patterns precede or follow the manifestation of HSAM, raising critical questions to address in future research about the causality and developmental trajectory of this superior memory ability.Given our finding of enhanced connectivity within midline cortical structures, we predict that behavioral interventions aimed at improving autobiographical memory in control participants -for instance, through structured memory training or narrative encoding strategies-could lead to measurable changes in the DN connectivity patterns akin to those observed in HSAM individuals.Such a finding would suggest that the distinctive network topologies we identified could be, at least partially, modifiable through targeted cognitive activities.Testing this prediction would not only further our understanding of the findings presented here, but would also potentially unveil mechanisms through which the extraordinary memory capacities of HSAM individuals are supported, thus possibly offering suggestions for interventions in populations with memory deficits (cf.Santangelo et al., 2022).
Finally, it should be noted that other potential factors related to more general cognitive differences could contribute to the observed neurobiological differences between the two groups.The current literature does not support the notion that individuals with HSAM have different cognitive abilities apart from their autobiographical memory abilities.Studies that focus on cognitive assessment of individuals with HSAM, such as LePort et al. ( 2017), do not provide evidence of superior psychometric intelligence in this group.Complementary findings show that HSAM individuals do not outperform controls on standardized tests (Patihis, 2016) or creative ability (Daviddi et al., 2022a), and are equally susceptible to false memories (Patihis et al., 2013).Regardless, a more comprehensive cognitive assessment could help to rule out potential confounding factors in future research.
In sum, our investigation provides insights into the network topologies underpinning HSAM, highlighting enhanced connectivity within midline cortical structures in a group of rare individuals with HSAM.At the same time, the current findings confirm a reduced resting-state MTL/saliency network functional connectivity as a potential neurobiological marker of HSAM.Overall, this study contributes to the growing body of literature exploring the boundaries of human memory capability by describing the brain network organization in individuals with HSAM.Our findings serve to inspire future lines of inquiry, such as the plasticity of the default network in response to environmental factors and potential applications of this knowledge to enhance memory function in clinical populations.Continued investigation of this phenomenon will not only illuminate the idiosyncrasies of HSAM but offer broader insights into the nature of human memory.

Data
The conditions our ethics approval do not permit public archiving of data when -also in an anonymized form-it is not possible to rule out any link to the individual's identity.Here we reported data belonging to a rare population of individuals that released several interviews to national and international press.For this reason, data anonymization cannot be fully guaranteed.Access will only be granted to named individuals in accordance with ethical procedures governing the reuse of sensitive data, upon completion of a data sharing agreement.Analysis codes and scripts are available on the publicly accessible digital repository 'Open Science Framework' (OSF; https://osf.io/4eb7v/).

Fig. 1 e
Fig. 1 e Detection of the Autobiographical Memory Network.(A) We perform complementary seed-based and whole-brain analysis to assess how specific cortical regions within the autobiographical memory network are differentially connected in HSAM individuals, and identify differences in network topology associated with enhanced autobiographical memory; (B)We identified five ROIs associated with autobiographical memory, which were used to generate seed-based connectivity maps.These ROIs include: medial prefrontal cortex (mPFC), posterior cingulate cortex (PCC), medial temporal lobe (MTL), angular gyrus (AG), and temporal pole (TP); (C) Participants underwent resting-state fMRI scans, from which we computed voxel-wise connectivity matrices between all gray matter voxels.We then summed all the weighted connections of each voxel to generate WD maps showing the extent to which each voxel is functionally connected to the rest of the brain.

Table 1 e
Regions of Interest.ROI refers to the contiguous cluster of voxels associated with autobiographical memory, defined via Neurosynth meta-analysis.Number of Voxels indicates the size of the ROI.Peak activation (MNI coordinates) provides the voxel within this ROI with the highest z-statistic.Peak activation (z-statistic) provides the value of the z-statistic for the corresponding voxel.

Table 2 e
Seed-Based Results.Seed refers to the region of interest previously defined in Table1.Cluster refers to the location of the significant voxels sharing connectivity to the seed.The labels for these clusters include the following abbreviations: Sign refers to the direction of this relationship.Number of voxels indicates the number of voxels contained in the cluster.Peak activation (MNI coordinates) provides the MNI coordinates for the voxel with the highest t-statistic within the cluster.Peak activation (t-statistic) provides the t-statistic for the corresponding voxel.