The role of fMRI in the mind decoding process in adults: a systematic review

Research question and hypothesis

This systematic review employs the PICO framework to formulate a research question regarding the role of fMRI in understanding adult mind-decoding (Schardt et al., 2007). It is defined as follows: Population (P): Adults in general. Interventions/Exposure (I): Utilization of fMRI techniques during mind decoding tasks. Comparison (C): No specific comparison is required for this research question, as the focus is on understanding the role of fMRI in the mind-decoding process. Outcome (O): Neural mechanisms and brain regions involved in the mind-decoding process.

With the PICO framework, our research question and hypothesis can be formulated as follows:

Research Question: How can fMRI help us understand the neural mechanisms and brain regions involved in mind-decoding tasks in adults?

Hypothesis: Employing fMRI in mind-decoding research will reveal specific neural networks and brain regions associated with this cognitive function in adults.

Search strategy

We used three online databases, including Web of Science (2004-2024), Google Scholar (2000-2024), and PubMed (2009–2024), to identify studies reporting the role of fMRI in mind decoding. We used the following search terms: “fMRI” OR “mind reading/decoding” OR “Neurodevelopmental changes” OR “emotional changes” OR “Social Cognition” OR “Mental State” AND “fMRI”. Relevant literature was identified and preserved for subsequent review in order to extract significant information. Only articles published in English were included in the search, and their adherence to the specified inclusion and exclusion criteria was assessed. The online search was conducted by four co-authors between February 1 and March 31, 2024. The search strategy was independently performed by Maher Mohammed Alotaibi and Faisal Saleh Alghamdi. Any disagreements during the screening process were resolved through discussion. If consensus could not be reached, Sahal Alotaibi and Hamid Osman acted as referees to make the final decision. Afterward, all co-authors collaborated to merge the results and remove any duplicates.

Selection criteria

The title, abstract, and full text of identified articles were screened independently by all authors to evaluate their eligibility for inclusion in this systematic review. Only studies that met all inclusion criteria were included, and the references cited within those studies were assessed for eligibility. The full texts of the selected studies were downloaded to facilitate thorough decoding and extraction of information, aligning with the primary purpose and objective of this study. Each record was independently screened by Mishaal Abdullah Alshehri, Khaled Majed Bamusa and Ziyad Faiz Almalki, and Sultan Alamri and Ahmad Joman Alghamdi to ensure thorough and unbiased evaluation. Again, any disagreements during the selection process were resolved through discussion. If consensus could not be reached, Sahal Alotaibi and Hamid Osman acted as referees to make the final decision.

Inclusion and exclusion criteria

To ensure high-quality reporting, we followed the guidelines specified in the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) (Page et al., 2021). Furthermore, the PRISMA flow chart was utilized to guide the selection process (see Fig. 1).

In this systemic review, because we intend to analyze the role of fMRI in the mind-decoding process, only studies meeting the following criteria were included: (a) the study included only adults; (b) studies published in an open-access or subscription journal; (c) studies published in English language only; (d) studies published on or after 2000; (e) studies that utilize fMRI as a neuroimaging modality; (f) studies that investigate mind decoding or mind-decoding related processes, such as mental state decoding, brain-computer interfaces, or neural correlates of mental processes; (g) studies employing various mind decoding techniques, such as pattern classification, multivariate analysis, or decoding algorithms.

The excluded studies: (a) studies not utilizing fMRI as a neuroimaging modality; (b) studies that do not focus on mind decoding or mind-decoding related processes; (c) studies involving non-human subjects; (d) studies published in languages other than English; (e) review articles, opinion pieces, editorials, and conference abstracts; (f) studies that primarily focus on other neuroimaging techniques or modalities, without substantial emphasis on fMRI and mind decoding.

Data extraction

The authors conducted an independent screening of the titles and abstracts of the identified articles. Subsequently, full-text articles were thoroughly reviewed to assess their eligibility for inclusion. Data extraction encompassed various aspects of the selected studies, including fundamental characteristics such as authorship, publication date, research type, and methods employed (including fMRI imaging parameters and experimental tasks). Additionally, the essential characteristics of the research subjects, such as age categories (specifically adults), were also extracted.

Quality assessment

Each of the authors independently performed a comprehensive quality assessment of the included studies, considering significant factors such as experimental design, statistical methods, reporting quality, methodological rigor, age group categorization, potential sources of bias, and the generalizability of findings.

Study selection

A total of 366 articles were initially identified from multiple databases (Web of Science, PubMed, and Google Scholar). After removing duplicates and screening titles and abstracts and paid access, 45 articles were selected for full-text review. Following the full-text review, as a result, a total of 10 studies met the eligibility criteria and were included in this systematic review.

Characteristics of studies

The included studies were published between 2000 and 2024, covering recent research in the field. All studies focused on adult populations, with participants ranging in age from 18 to 60 years. Various study designs were represented, including cross-sectional, experimental, and longitudinal studies. The studies utilized fMRI as the primary neuroimaging modality to investigate mind decoding or related processes.

Summary of results related to the role of fMRI in the mind decoding process

The selected studies provided important insights into the topic. (A) One study used the reading & decoding mind in the eyes test (RME) to assess mind decoding among typically developed individuals (Kawata et al., 2012). The study calculated differences in Blood-oxygen-level dependent (BOLD) signals between two tasks (A and B) using paired t-tests. The results revealed significant differences in the activation of the right superior occipital gyrus and left parietal lobe, suggesting that these brain regions are specifically involved in mind decoding. These findings were consistent with previous MRI studies using the RME.

(B) Another study focusing on decoding narratives and emotions with fMRI (Wang et al., 2011); the main objective was to explore the relationship between anxiety disorders and brain activation during story decoding. The study involved 14 participants with anxiety disorders and 14 control participants. A notable finding was the identification of a significant interaction between the group (anxiety disorder vs. control) and emotion in a cluster of brain activation located in the thalamus. Specifically, participants with anxiety disorders exhibited significantly lower levels of thalamic activation compared to the control group. This interaction was detected by comparing brain responses to emotional passages versus neutral passages.

(C) This study examined the electrophysiological correlates of decoding the single mind and decoding the interactive mind using Chinese idioms (Toazza et al., 2022). The study focused on the activation levels of the medial prefrontal cortex (MPFC), temporal lobe, precuneus, and occipital lobe. The results indicated that these brain regions showed increased activation when stimuli involved two persons compared to stimuli with only one person. These findings aligned with previous fMRI results and event-related potentials (ERPs) studies. Additionally, task-based fMRI studies have reported fMRI signal activities that are linked to specific task; speech sound (Alotaibi et al., 2023) and eye movements (Aloufi, Rowe & Meyer, 2021) (see Figs. 2 and 3).

(D) In the fMRI study of Bliksted et al. (2019), researchers investigated social cognition in 17 individuals recently diagnosed with first-episode schizophrenia (FES) who had limited or no exposure to antipsychotic medication. These patients were compared to a group of healthy controls matched on a 1:1 basis. The fMRI results revealed that in the control group, intentional movement associated with theory of mind (ToM) activation led to greater activation in specific brain regions, including the temporal gyrus, occipital cortex, and inferior frontal cortex. This activation pattern aligned with previous research findings that observed a response in the posterior superior temporal sulcus (pSTS) for this particular contrast, validating the expected normal response. Interestingly, the patients with first-episode schizophrenia exhibited a qualitatively similar pattern of activation in these brain regions. However, when comparing the control and patient groups, a distinction was identified in another region, which is further elaborated in the supplementary material (Bliksted et al., 2019).

(E) In the study investigating the effects of aging on mind decoding using fMRI (Castelli et al., 2010), two groups of healthy adults were included: a young group with an average age of 25.2 years and an old group with an average age of 65.2 years. Participants from both groups underwent fMRI scanning while performing the Decoding the Mind in the Eyes test, which involves attributing mental states to others based solely on their eye expressions. Interestingly, there were no differences in behavioral performance between the young and old groups. Additionally, both groups exhibited activation in the posterior superior temporal sulcus (pSTS) and the temporoparietal junction (TP), indicating that older individuals do not display impairments in the circuits associated with mentalizing processes. (F) In this ongoing study (Gómez-Carmona et al., 2021), the primary aim was to examine the brain regions that are activated when individuals process dishes with a pleasant design compared to those with an unpleasant design. Additionally, the researchers investigated how previously read rational descriptions compared to emotional descriptions influenced the visualization of the dishes. The study utilized fMRI to analyze brain activity. The findings indicated that participants who visualized pleasant dishes exhibited activation in various domains, including attention, cognition, and reward. Conversely, visualizing unpleasant dishes resulted in stronger activation in regions associated with inhibition, rejection, and ambiguity.

Interestingly, when participants read rational descriptions while visualizing pleasant dishes, brain regions linked to congruence integration were activated. In contrast, participants who visualized emotional descriptions exhibited an increased neuronal response to pleasant dishes in regions associated with memory, emotion, and congruence (Gómez-Carmona et al., 2021). (G) In Underwood et al. (2021), the primary aim was to investigate the neural circuits involved in appraising threatening and non-threatening situations in individuals with and without a need for care for psychotic experiences (PEs). The study included three groups: patients with psychosis spectrum disorder, non-need-for-care participants with PEs, and healthy control participants without PEs. fMRI was used while participants completed the Telepath task, which induced anomalous perceptual experiences. The researchers examined perceived threat evaluations and brain responses during the task, looking for group differences and correlations between brain activation and threat appraisals. The clinical group reported higher subjective threat appraisals compared to both the non-clinical and control groups, while there were no differences between the two non-clinical groups. Additionally, the clinical group exhibited lower BOLD signals in the superior and inferior frontal gyri compared to the non-clinical and control groups. Activation in the precuneus was found to correlate with threat appraisals reported during the task. The findings suggest that intact functioning of fronto-parietal regions may contribute to resilience and the ability to contextualize and evaluate psychotic experiences in individuals with persistent anomalous experiences. (H) In Pirastru et al. (2023), 62 healthy subjects were divided into two groups: group A, consisting of 21 subjects who underwent test-retest reliability assessments on the same day using the same task form, and group B, consisting of 30 subjects who underwent test-retest reproducibility assessments with a 4-month interval using two equivalent-parallel task forms. The results showed that in the voxel-wise general linear model (GLM) analysis, there were no significant differences in reliability and reproducibility for both conditions.

However, when using a region of interest (ROI)-based approach and considering areas with significant main effects of the stimuli, the reliability, as measured by the intraclass correlation coefficient (ICC), was poor (<0.4) for the positive condition and ranged from poor to excellent (0.4–0.75) for the negative condition. The ICC-based reproducibility analysis, which compared two different parallel task forms, yielded similar results. (I) In Reddy, Tsuchiya & Serre (2010) (RME decoding category), participants completed seven or eight fMRI scanning runs, each lasting approximately 5.6 min. Each run included fixation blocks, visual presentation blocks, and visual imagery blocks. During the visual presentation blocks, participants were shown four categories of objects (food, tools, famous faces, and famous buildings) in separate blocks. Each category was presented twice per run, with four exemplars per category. Each trial within a block consisted of 2 s of visual presentation and 2 s for task response, and the trial order was randomized. The study involved two fMRI experimental conditions: visual presentation (P) and visual imagery (I), both of which were completed by the participants. In the P condition, they viewed exemplars from four stimulus categories (food, tools, famous faces, and famous buildings) in separate blocks. In the I condition; participants heard auditory instructions and were asked to imagine the stimuli. Different exemplars were used in different runs. Prior to the scans, participants familiarized themselves with the stimuli. Activation patterns in the imagery condition were smaller compared to the perception condition (Reddy, Tsuchiya & Serre, 2010). (J) In Tamir et al. (2016), a total of 26 participants took part (10 male, 16 female) who were right-handed, native English speakers without any neurological issues. The participants’ average age was 21.2 years, ranging from 19 to 26 years. Prior to participation, all individuals provided consent following the guidelines approved by the Committee on the Use of Human Subjects at Harvard University. In this investigation, we performed initial analyses to pinpoint the brain regions that exhibited responses to two specific characteristics: (i) the vividness of passages describing physical scenes and events, and (ii) whether the passages described a person or a person’s mental content. Consistent with previous research, both vivid passages and passages with social content showed increased activation in regions of the default network compared to abstract and non-social passages. Specifically, when comparing social versus non-social passages, we found activity in the dorsomedial prefrontal cortex (dmPFC), ventromedial prefrontal cortex (vmPFC), lateral temporal cortex (from the temporal pole to the temporoparietal junction), bilateral hippocampi, and bilateral inferior frontal gyrus (IFG). When comparing vivid versus abstract passages, we observed robust activity in the medial temporal lobe (MTL) structures, including the bilateral hippocampus and para-hippocampus, as well as the retro splenial cortex and precuneus (Tamir et al., 2016).

Overall, the studies provided valuable insights into the role of fMRI in understanding the mind-decoding process, highlighting specific brain regions and their involvement in various contexts. Individual group activations are further detailed in the Supplementary Information.

Limitations of mind-decoding techniques

• Decoding thoughts and intentions from brain activity using fMRI is still under development and may not always be accurate.

• The effectiveness of mind-decoding techniques using fMRI might be influenced by the specific tasks used in the studies. Techniques might not generalize to real-world scenarios.

• As fMRI technology advances, ethical concerns arise regarding privacy and potential misuse of mind-decoding capabilities.