Candidate Mechanisms for Naturalistic fMRI Superiority
(a) Signal Amplification and Ecological Specificity: Naturalistic fMRI paradigms (e.g., movie-watching or story-listening) provide a continuous, multimodal, and ecologically valid stream of sensory information that better approximates how the human brain processes the real world [1, 2]. During resting-state fMRI, brain state dynamics are largely dominated by bistable transitions between two relatively indistinct states [3, 4]. In contrast, naturalistic viewing reshapes these intrinsic dynamics into a richer, hierarchical repertoire of well-defined functional brain states [3, 5, 6]. Movie viewing induces deeper attractor networks, causing stronger temporal perturbations away from the global mean and creating a steeper local attractor landscape [6, 7]. This robustly amplifies higher-order neural processes and signal properties compared to the unconstrained resting state [8].
Better SNR and Controlled Cognitive State: Resting-state fMRI is heavily confounded by unconstrained mental activity (e.g., drifting wakefulness, daydreaming) and non-neuronal physiological noise such as head motion, respiration, and cardiac artifacts [9-11]. Naturalistic stimuli exert an implicit behavioral constraint that captures the participant's attention and reduces anxiety [1]. Crucially, this engagement translates to significantly reduced head motion—a primary source of artifact in fMRI, particularly in pediatric and clinical populations [12-14]. Furthermore, novel analytical methods like Inter-Subject Functional Correlation (ISFC) can be applied to naturalistic data to filter out intrinsic, idiosyncratic neural dynamics and non-neuronal noise, thereby massively increasing the signal-to-noise ratio (SNR) for detecting stimulus-induced network configurations [11, 15, 16].
Individual-Engagement Variance Amplification: Because naturalistic stimuli drive highly consistent neural responses across subjects (synchronizing the "baseline" perceptual processing) [17-19], the remaining idiosyncratic deviations in brain state dynamics become highly meaningful markers of individual differences [4, 20]. This phenomenon, known as "neural polarization," demonstrates that how an individual cognitively and emotionally interprets a narrative directly alters their high-order neural network activity [21]. Consequently, subject-specific idiosyncrasies in functional connectivity (FC) during movies heavily correlate with subjective appraisal, emotional engagement, and memory reinstatement [22-24].
(b) Disorders/Traits Benefiting Most and Effect-Size Comparisons
Naturalistic fMRI has shown profound advantages for specific clinical and trait-based predictions:
- Depression (MDD): Naturalistic stimuli can elicit symptoms and affective states that are subtle or absent during rest [8]. For example, minor depression is characterized by neural polarization and aberrant inter-subject correlations within the Default Mode Network (DMN), specifically the posterior cingulate cortex and precuneus, when viewing emotionally valenced films [25-27]. Patients with melancholia also show disrupted neural connectivity in emotional circuitry during sad film viewing [28, 29].
- Anxiety and Stress: Real-world cinematic stressors have been used to continuously track physiological arousal (e.g., heart rate) and map it to dynamic shifts in large-scale networks. During highly anxiogenic movies, cohesion within the salience network increases, mirroring the network-level aberrations observed in Generalized Anxiety Disorder (GAD) and Post-Traumatic Stress Disorder (PTSD) [30, 31].
- Autism Spectrum Disorder (ASD) & ADHD: Beyond the practical benefit of significantly reducing head motion in these groups [12-14], naturalistic viewing reveals highly idiosyncratic cortical activation patterns in individuals with ASD that deviate from neurotypical inter-subject synchronization [19].
- Personality and Traits: Trait paranoia, psychological resilience, intelligence, and internalizing/externalizing symptoms have been successfully mapped to unique functional connectivity signatures evoked by naturalistic narratives [26, 32-34].
Effect-Size Comparisons: When benchmarking naturalistic paradigms against resting-state and traditional task fMRI, the empirical gains in reliability and predictive power are substantial.
- Test-Retest Reliability: The reliability of functional connectivity and graph theoretical measures improves significantly during natural viewing compared to rest, exhibiting an average increase in intra-class correlation (ICC) of almost 50% across various connectivity measures [35-37].
- Classification Accuracy: When using ISFC to decode cognitive states or classify stimulus conditions, accuracy jumps dramatically compared to standard resting-state FC. In one study, classification accuracy between experimental conditions was 80% using ISFC during narrative processing, compared to just 37% using standard FC [38].
- Connectome Fingerprinting: Unsupervised algorithms designed to match an individual's FC matrix across different scanning sessions achieve much higher accuracies when utilizing movie-watching data compared to rest, indicating that complex tasks make individual differences in the connectome much more distinct [20].
(c) AI-Driven Adaptive Stimulus Selection for Personalized Diagnostics
The integration of artificial intelligence offers a transformative pathway to personalize naturalistic fMRI diagnostics. Because specific movie features (e.g., luminance, speech, facial expressions, and emotional valence) reliably drive discrete, predictable brain states [39, 40], stimuli can be engineered to target specific neural circuits [41].
Recent advancements have demonstrated that deep neural networks (DNNs) and large language models (LLMs) can map high-level semantic, categorical, and acoustic features of video and audio directly to human cortical processing hierarchies [42-45]. Furthermore, researchers have successfully used self-generated personal narratives, obtained through interviews, as fMRI stimuli to evoke cognitive and emotional states that perfectly mimic a patient's spontaneous thoughts (e.g., rumination or daydreaming) [46-48]. By building predictive models based on how the brain responds to these personalized stories, researchers can decode the self-relevance and affective valence of a patient's internal mental state [49, 50].
Synthesizing these approaches, AI could be utilized to dynamically curate, edit, or generate naturalistic stimuli—such as LLM-generated narratives tailored to a patient's specific depressive ruminations or trauma, or dynamically edited video sequences designed to provoke targeted social cognition deficits in ASD [41, 48]. By moving away from "one-size-fits-all" commercial movies and instead using AI to generate computationally optimized, patient-specific continuous stimuli, clinicians could deliberately perturb vulnerable neural circuits, thereby maximizing the individual biomarker signal and the accuracy of psychiatric diagnoses [40, 41, 48].
[1] (src:bffb7be6) Recently, the use of naturalistic stimuli, such as movies and music, is gaining increasing traction in cognitive neu-roscience [Hasson and Honey, 2012; Spiers and Maguire, 2007]. These naturalistic paradigms have provided novel insights on how human brain functions in real-life context, which is mor...
[2] (src:43b9cbf0) 1. Introduction The term “naturalistic paradigms” traces its roots to a shift in the field of vision research from using highly constrained artificial stimuli such as bars of light to also studying neural responses to complex nat-ural images such as landscape photographs (for reviews, see (Felsen an...
[3] (src:d830a9d5) show that the temporal dynamics of brain states, as measured in fMRI, are reshaped from predominantly bistable transitions between two relatively indistinct states at rest, toward a sequence of well-defined functional states during movie viewing whose transitions are temporally aligned to specific f...
[4] (src:d830a9d5) We also assessed the inter-session consistency by calculating the Jaccard index over state visits across session A and session B, averaged over brain states and participants (see Methods). The occurrence of brain states was significantly more consistent during movie viewing than rest (average Jaccar...
[5] (src:d830a9d5) Brain states occurring at rest and during movie viewing were estimated using the HMM, a method which posits that the observed data arise from a small number of hidden states and their transitions5. To allow for a direct comparison between the states during the different experimental conditions (base...
[6] (src:d830a9d5) (milliseconds39,45) resting-state dynamics that reflect genetic and behavioural traits, including intelligence5. The current work demonstrates that this generative modelling technique can cap-ture subject-specific brain state transitions that are temporally aligned to perceptual, semantic, and narra...
[7] (src:40b3e36e) increasedmagnitude. This indicates that theneural activitymoreeasily fell into the DMN attractor because it was in a steeper landscape when attentive. This is in line with studies that reported high activity in DMN atmoments of optimal performance during this task45–47. These results highlight the f...
[8] (src:864419ff) The naturalistic paradigm combined with fMRI shows substan-tial potential in advancing research on emotional dysregulation of psychiatric disorders. For example, movie-fMRI, employing films with feature-rich content, can robustly induce genuine mood states (Hasson et al. 2008), enhance higher-order ...
[9] (src:bffb7be6) The majority of research on functional connectivity net-works has been conducted with resting-state functional magnetic resonance imaging (fMRI) paradigms. With low performance demand and high compliance, resting-state fMRI hence minimizes behavioral confounds normally pre-senting during task condit...
[10] (src:d830a9d5) The variability of resting-state neural dynamics across parti-cipants and their unconstrained nature limits the ability to make direct inferences about the behavioural relevance of brain state dynamics. Furthermore, common methods adopted to infer macroscopic dynamics, including dynamic functional c...
[11] (src:7b279216) To better characterize the dynamic changes in DMN correlation patterns that are locked to the processing of external stimuli, we introduce a novel method termed inter-subject functional correlation (ISFC), in which inter-region correlations are calculated between different brains exposed to the same...
[12] (src:43b9cbf0) Despite these challenges, a major reason for the growing use of naturalistic paradigms in developmental fMRI has been the effect of movie-watching on head motion during scanning. With conventional task-based fMRI, scanning young awake children was possible because child-friendly tasks maintained int...
[13] (src:43b9cbf0) The idea of using movies to improve compliance is not new; labs and hospitals have been showing cartoons during structural MRI se-quences to improve image quality and avoid sedation during pediatric scans for years (Raschle et al., 2012, 2009). In 2013, Cantlon and Li published an elegant study usin...
[14] (src:43b9cbf0) (20.3 ± 6.3) with Down Syndrome and healthy participants ReHo= regional homogeneity, ALFF= amplitude of low frequency fluctuations, FC= functional connectivity, ISCs= intersubject correlations, ToM=Theory of Mind, GLM=general linear model, DAN=dorsal attention network, ICA= independent component ana...
[15] (src:7b279216) In a task setting, the pattern of correlations within each individual, as computed by the FC method, will be influenced by each of the three components (Fig. 1b). In contrast, inter-subject correlation (ISC)26 captures the stimulus-induced correlation across subjects within a given region by correla...
[16] (src:7b279216) the precuneus and the insula, across two groups, each of 18 subjects, during intact story condition. Discussion To study the DMN’s functional properties, one needs to investigate its responses while it processes information from the environment, that is, during task performance. However, this is a c...
[17] (src:d830a9d5) States are consistent over participants during movie watching. We next assessed the degree of inter-subject consistency in the expression of these ten states while participants watched the movie or underwent the resting-state condition. Using a reverse-inference approach3,24, we investigated whether...
[18] (src:d830a9d5) As anticipated, movie viewing was associated with greater consistency across participants, relative to rest (Fig. 2; also Supplementary Figs 3, 4, and 6). Moreover, the co-occurrence of brain states across participants reached the highest levels during specific movie events (Table 1), with complete ...
[19] (src:43b9cbf0) (i.e., time-locked) signal changes, but are also model-free and whole-brain. ISCs can be examined in groups of different developmental stages (see Fig. 5), or across clinical groups such as individuals with autism spectrum disorder (Bolton et al., 2018; Hasson et al., 2009) or Down Syndrome and heal...
[20] (src:43b9cbf0) 13.3. Individually distinct patterns of FC Recent work has investigated individually distinct patterns of FC by employing an unsupervised test-retest matching algorithm to identify individual subjects from within a group based solely on the correlation strength between FC matrices (Finn et al., 2015...
[21] (src:864419ff) Based on the ISC, Leong et al. (2020) introduced the con-cept of neural polarization, which refers to the neural activity shared between individuals with similar political attitudes but not between individuals with dissimilar political attitudes. Neu-ral polarization constitutes the neurobiological ...
[22] (src:d830a9d5) Between subject differences in dynamics link to movie ratings. We then investigated if brain state dynamics unique to each participant were associated with their subjective ratings of the movie. Subjective ratings were obtained using a simple ques-tionnaire containing questions about (i) boredom, (i...
[23] (src:d830a9d5) into increased expressions of brain state 1, 7, and 8 combined with reduced expressions of brain states 2 and 3, and vice versa. This finding is associated with more transitions from state 3 to state 4. These results appeared to be specific to movie viewing as brain state dynamics extracted from the...
[24] (src:d830a9d5) physiological changes and brain state dynamics suggest that the sensory properties of a movie, as well as the content of its nar-rative, could be manipulated to evoke discrete brain processes. These findings motivate further investigations into the use of structured naturalistic stimuli to induce se...
[25] (src:864419ff) Abstract Research on the neuropathological mechanisms underlying minor depression (MD), particularly in individuals with a history of recur-rent minor depressive episodes, is very limited. This study focuses on the abnormality in processing real-life emotional stimuli among individuals with MD. Thir...
[26] (src:864419ff) In this study, we adopted movie-fMRI to investigate the neu-ral polarization among individuals with MD and normal controls (NC). Of note, we focused on the emotion processing of individ-uals with MD who experienced recurrent depressive episodes in recent years. Thirty-two participants with MD and 31...
[27] (src:864419ff) The mediating role of neural similarity As shown in Fig. 4, neural similarities to NC vs. MD of the PCC and PCUN were negatively correlated with depression severity (q’s < .05). Under the negative and neutral emotional conditions, the neural similarity to NC vs. MD of the PCC and PCUN mediated the r...
[28] (src:bffb7be6) dynamics and connectivity during natural movie viewing in autism, major depressive disorder, and altered states of consciousness [Guo et al., 2015; Hasson et al., 2009; Hyett et al., 2015; Naci et al., 2014]. Therefore, naturalistic para-digms could provide a promising condition for mapping connecti...
[29] (src:864419ff) With the emergence of the naturalistic paradigm, two methods have been proposed: inter-subject correlation (ISC) and inter-subject functional connectivity (ISFC). The ISC analysis is com-monly used to measure the shared neural response across sub-jects under conditions with complex information (Hass...
[30] (src:6d522330) Received May 31, 2016; revised Sept. 13, 2016; accepted Sept. 16, 2016. Author contributions: I.T., G.F., and E.J.H. designed research; D.E. performed research; C.B.Y., G.R., C.F.B., T.H., This work was supported by the National Science Foundation (DGE0824162 to C.B.Y.), the Netherlands Organi-satio...
[31] (src:6d522330) Implications for psychopathology Importantly, our results extend a prominent triple network model of psychopathology (Menon, 2011) by suggesting that stress-related mental disorders involve inadequate network switching due to frequent or constant high-arousal states. Similar to our findings of incre...
[32] (src:864419ff) Furthermore, ISC is growing in popularity as a window into the neural correlates of stable ability or trait (Finn et al. 2018, Jangraw et al. 2023), thus promising for exploring an individual’s intrinsic qualities. Psychological resilience, mediated by the adaptation in brain functional circuits, is...
[33] (src:43b9cbf0) Felsen, G., Dan, Y., 2005. A natural approach to studying vision. Nat. Neurosci. 8 (12), 1643–1646. Finn, E.S., et al., 2015. Functional connectome fingerprinting: identifying individuals using patterns of brain connectivity. Nat. Neurosci. 18 (11), 1664–1671. Finn, E.S., et al., 2017. Can brain sta...
[34] (src:0334945f) Brain Androgyny Associated with Fewer Internalizing Symptoms In the HCP cohort, we found that the internalizing score, but not the externalizing score, was associated with the second-order term of the brain gender continuum (df = 684; r = 0.08; p.perm = 0.0409,uncorrected; Fig. 4A). This U-shaped re...
[35] (src:bffb7be6) Test–Retest Reliability of Functional Connectivity Networks During Naturalistic fMRI Paradigms Jiahui Wang,1 Yudan Ren,1 Xintao Hu,1 Vinh Thai Nguyen,2 Lei Guo,1 Junwei Han,1* and Christine Cong Guo2* 1School of Automation, Northwestern Polytechnical University, Xi’an, China 2QIMR Berghofer Medical ...
[36] (src:901d683f) challenging populations such as children or cognitively impaired patients (Kim, Wang, Wedell, & Shinkareva, 2016; Kuo et al., 2019; Mandelkow, de Zwart, & Duyn, 2016). To further validate its poten- tial in functional brain network studies, the elucidation of test–retest reliability of FC in natural...
[37] (src:901d683f) improved dFC-ICC compared to resting state though the effect of scan order may exist. 5 | CONCLUSION Utilizing two scan sessions of resting-state and natural viewing fMRI data from the same group of subjects, the test–retest reliabil- ity of dFC statistics were investigated and compared between the ...
[38] (src:7b279216) Across-subject classification of stimulus type. Next, we trained a classifier to quantify the improvement in discriminating between the four experimental conditions by using ISFC over FC. Clas-sification was performed separately using ISFC and using FC (see Supplementary Note 2). Classification accu...
[39] (src:d830a9d5) 2 (high DMN, salience but low executive). In line with high DMN activity, state 2 is functionally linked to high anxiety29 and pain30. PD may therefore also link, at least to some extent, to transient interoceptive mechanisms. Previous work has shown evoked PD in the absence of visual stimuli may re...
[40] (src:43b9cbf0) Every movie contains numerous visual and auditory features such as luminance change and the zero-crossing rate of the soundtrack, the complexity of which refers back to a common concern regarding nat-uralistic paradigms as mentioned in the Introduction. Each of these features can be used as a regres...
[41] (src:43b9cbf0) 13.5. Paradigms to evoke specific symptom domains To date, researchers have selected movies based on availability, personal preference, the intention to elicit strong emotional responses, or to be more neutral. As an example of the latter intent, Inscapes was created to evoke a specific type of proc...
[42] (src:67a91294) A unique challenge for data analysis posed by brain recordings during naturalistic tasks (e.g. movie-watching) is the presence of linearly and nonlinearly correlated confounding variables. These limit the effectiveness of standard statistical tools such as t-tests. However, encoding models may prese...
[43] (src:b0437311) After data preprocessing, we applied multivariate pattern analysis and extracted the neural representations of the stimuli using representational similarity analysis107 (Fig. 1b; see Methods). For fMRI, we obtained the evoked response patterns for each stimulus at either a region of interest (ROI) o...
[44] (src:b0437311) To examine howdifferent levels of stimuli information are represented in the brain, we used different computational models to capture low-level visual (GIST descriptors109) and acoustic features (Cochleagram model110) and high-level categorical and semantic features (GloVe word embeddings111) (Fig. ...
[45] (src:b0437311) representations and achieve higher task performance113. To evaluate the hierarchical correspondence between the DNN model and the brain, we extracted model activations from seven blocks of each model branch and used them to construct RDMs. We observed that representations in early layers of video an...
[46] (src:16cdf350) personal story | spontaneous thought | functional magnetic resonance imaging | brain decoding | affective neuroscience Our mind never rests. Even during quiet periods or sleep, our mind spontaneously wanders from the past to the future and from one concept to another (1–3). Spontaneous thoughts may...
[47] (src:16cdf350) As shown in Fig. 1A, the major research goals of the current study include 1) developing fMRI multivariate pattern- based predictive models of self- relevance and valence using data from the story- reading task, 2) comparing and interpreting the newly developed predictive models of self- relevance a...
[48] (src:16cdf350) First, we used personal narratives as experimental stimuli to induce cognitive and emotional states similar to spontaneous thoughts. We used self- generated personal stories as stimuli for the following reasons: 1) self- relevant thoughts, such as current personal concerns, past memories, and future...
[49] (src:16cdf350) The contents and dynamics of spontaneous thought are important factors for personality traits and mental health. However, assessing spontaneous thoughts is challenging due to their unconstrained nature, and directing participants’ attention to report their thoughts may fundamentally alter them. Here...
[50] (src:16cdf350) Discussion In this study, we developed multivariate pattern- based predictive models of self- relevance and valence that can be used to decode affective dimensions of spontaneous thoughts. For this, we con-ducted an fMRI experiment using a narrative- reading task, in which we showed personal stories...
Sources used: 12 documents
- bffb7be6-4fe1-474a-a013-e95e211e712e
- 43b9cbf0-9356-4700-9922-64658c6fff03
- d830a9d5-a8b2-4a71-9182-b901f9aacbc9
- 40b3e36e-af9e-4c96-9f74-fc80f3a5bb5f
- 864419ff-95ee-4a93-a58f-2e1c480fce22
- 7b279216-6743-406c-9009-741ebfeed269
- 6d522330-79db-4efd-8f8d-8d2b94dc614e
- 0334945f-57d7-427f-9ff6-12c33826de6a
- 901d683f-e842-4bf4-9441-93b858d20e07
- 67a91294-ea55-4ee2-b94f-1676af02528a