Informed consent competency assessment for brain-computer interface clinical research and application in psychiatric disorders: A systematic review

Abstract

BACKGROUND

Brain-computer interface (BCI) technology is rapidly advancing in psychiatry. Informed consent competency (ICC) assessment among psychiatric patients is a pivotal concern in clinical research.

AIM

To analyze the assessment of ICC and form a framework with multi-dimensional elements involved in ICC of BCI clinical research among psychiatric disorders.

METHODS

A systematic review of studies regarding ICC assessments of BCI clinical research in patients with six kinds of psychiatric disorders was conducted. A systematic literature search was performed using PubMed, ScienceDirect, and Web of Science. Peer-reviewed articles and full-text studies were included in the analysis. There were no date restrictions, and all studies published up to February 27, 2025, were included.

RESULTS

A total of 103 studies were selected for this review. Fifty-eight studies included ICC factors, and forty-five were classified in ICC related ethical issues of BCI research in six kinds of psychiatric disorders. Executive function impairment is widely recognized as the most significant factor impacting ICC, and processing speed deficits are observed in schizophrenia, mood disorders, and Alzheimer’s disease. Memory dysfunction, particularly episodic and working memory, contributes to compromised ICC. Five core ethical issues in BCI research should be addressed: BCI specificity, vulnerability, autonomy, dynamic ICC, comprehensiveness, and uncertainty.

CONCLUSION

A Five-Dimensional evaluative framework, including clinical, ethical, sociocultural, legal, and procedural dimensions, is constructed and proposed for future ICC research in BCI clinical research involving psychiatric disorders.

Key Words: Informed consent competency; Brain-computer interface; Psychiatric disorders; Decision-making capacity; MacArthur Competence Assessment Tool

Core Tip: Informed consent among psychiatric patients is a major concern in brain-computer interface (BCI) clinical research. This review identifies informed consent competency (ICC) assessment as the main challenge, particularly in patients with schizophrenia, mood disorders, anorexia nervosa, alcohol dependence, and neurocognitive disorders due to Alzheimer’s disease and Parkinson’s disease. The MacArthur Competence Assessment Tool evaluates four clinical aspects of ICC. Six psychiatric disorders have common and disorder-specific factors and ethical issues related to ICC. The Five-Dimension BCI-specific ICC framework is constructed in psychiatric disorders. Future research should develop dynamic evaluation systems and enhance ICC of psychiatric patients in BCI clinical studies.

INTRODUCTION

The development of brain-computer interface (BCI) technology has significantly advanced our understanding of brain function, offering new insights into neural activity and the diagnosis and treatment of psychiatric disorders, and has seen increasingly widespread application. BCI technology is a new interdisciplinary technology that creates an informational channel between the brain and an external device, enabling direct interaction between the two[1]. Unlike conventional psychiatric treatments, which primarily rely on medicine, psychotherapy, physical therapy, and so on, BCI interacts directly with neural activity, providing a more targeted approach to brain function modulation. BCI collects neural activities with non-invasive or invasive methods through recording devices, decodes these activities using machine learning models, analyzes the contained subjective intentions and other information, outputs corresponding commands based on this information, manipulates external devices to perform actions consistent with human wishes, and receives feedback signals from these devices, creating an interactive closed-loop system[2]. With highly diverse systems, BCI varies in invasiveness and signal processing methods, resulting in different clinical applications. Invasive BCIs require a neurosurgical procedure to place recording sensors at the very source of the generation of neural activity to collect precise neural signals. In contrast, non-invasive BCIs use external devices to collect brain activity but lack precise task-related neuronal signals. An intermediary approach is a semi-invasive BCI, which requires a craniotomy but does not involve sensors penetrating the nervous tissue[3]. This shows that BCI is far more technically complex and presents unique challenges due to heightened medical, ethical, and technical risks, and informed consent related to BCI research and applications has become one of the most important issues.

Informed consent is built upon the elements of information, decisional capacity, and voluntarism[4]. Information on the consent process generally encompasses core elements, not only scientific, cultural, and social factors that are particularly relevant to designing ethically responsible approaches to informed consent, but also the importance of continued development of new, effective approaches to the process of informed consent[5]. Decision-making capacity (DMC) in clinical settings refers to an individual’s ability to understand relevant information and make informed choices within the context of research or clinical application. The classic DMC model comprises four key elements: Choice, understanding, appreciation, and reasoning[6]. Choice involves the ability to make and communicate a clear decision. Understanding is the capacity to grasp the information relevant to the decision, while appreciation refers to recognizing how this information applies to one’s own situation. Finally, reasoning is the ability to logically evaluate the information to make a sound decision[7]. Competency is an important presupposition to autonomous decision-making[8]. Assessing informed consent competency (ICC) becomes increasingly critical when the patients are psychiatrically vulnerable as potential participants of an innovative BCI research or application.

To date, few systematic reviews have addressed the intersection of ICC and BCI research in psychiatric disorders. Conducting BCI clinical research in the context of psychiatric disorders involves unique clinical risks, ethical challenges, legal, social, and cultural elements, making ICC particularly important. We conducted a systematic review of the current methods for assessing ICC, its influence, and associated factors related to ICC in BCI clinical research and its application in psychiatric disorders, as well as future research directions. We further discussed the ethical challenges and disorder-specific complexities of ICC from a bioethical perspective, aiming to provide a scientific foundation for developing novel ICC assessment technologies and methods in this cutting-edge field.

MATERIALS AND METHODS

Search strategies and selection criteria

This systematic review was carried out following the preferred reporting for systematic reviews and PRISMA guidelines[9]. A comprehensive auto electronic search was carried out on February 27, 2025, of the following three major databases: PubMed, ScienceDirect, and Web of Science. The search strategy employed a combination of controlled vocabulary and free-text terms to ensure the inclusion of all relevant literature. The search string was: (Brain computer interface OR BCI) AND (informed consent competency OR decision making capacity OR ethical issue) AND (psychiatry OR neuroscience OR schizophrenia OR mood disorders OR anorexia nervosa OR alcohol dependence OR major/mild neurocognitive disorder due to Alzheimer’s disease, OR Parkinson’s disease). To be included in this review, studies had to meet the following criteria: (1) Articles must be peer-reviewed journal articles with full texts available in English; (2) Studies must address issues related to ICC or DMC or ethical challenges within the context of BCI research; and (3) The participants targeted in this review were those with six major psychiatric disorders: Schizophrenia, mood disorders, anorexia nervosa (AN), alcohol dependence, major or mild neurocognitive disorder due to Alzheimer’s disease (AD), and Parkinson’s disease (PD). The following studies were excluded: (1) Editorials or conference abstracts; (2) Low-quality studies according to the Newcastle-Ottawa scale (NOS); (3) Articles focused exclusively on technical aspects such as algorithm development without discussing ICC concerns; and (4) Full text could not be obtained despite efforts to contact the corresponding authors. Titles and abstracts were screened to identify potentially relevant studies, and reference lists of the selected articles were manually reviewed to ensure the completeness of the search.

Data extraction and quality assessment

Data from the included articles were extracted using a standardized form to ensure consistency and facilitate qualitative synthesis. Data extraction was performed by two independent reviewers (Si JY and Lin ZY), who screened all titles and abstracts in duplicate and assessed the methodological quality of all the studies according to the NOS criteria[10]. The NOS, which utilizes a star-based rating system, allocates a maximum of 4 stars for the selection of study groups, 3 stars for the assessment of exposure or outcome, and 2 stars for the comparability of cohorts, yielding a total possible score of 9. In the present review, a minimum NOS score of 5 was established as the threshold for study inclusion[10]. In instances of disagreement between the two primary reviewers, a third reviewer (Wang XQ) was consulted to resolve discrepancies and achieve consensus. Extracted data included publication details (including author name and year of publication), issues discussed (such as informed consent or decision making), characteristics of the study patients (6 major psychiatric disorders), and references to ICC or DMC evaluation frameworks. Data synthesis used narrative synthesis[11] and tabular comparison[12]. The GRADE approach to rating the quality of evidence provides a framework for evaluating five factors that may affect evidence quality and was used in this systematic review: Risk of bias, imprecision, inconsistency, indirectness, and publication bias. It offers effect estimates for each outcome to support use by decision-makers, clinicians, and patients, thereby enabling transparent and explicit judgments[13].

This systematic review has been registered in the International Prospective Register of Systematic Reviews with the registration number CRD420251035965.

RESULTS

The search identified 62442 potentially eligible studies from the three databases: PubMed (n = 5188), Science Direct (n = 41336), and Web of Science (n = 15918). An additional 52 studies were included through manual searching. After removing 76 duplicates, 62418 records were sought for retrieval. Records were excluded with full text not available (n = 10195). A total of 52223 full-text articles were assessed for eligibility, from which 52120 were excluded as they were non-English (n = 151), irrelevant to ICC concerns (n = 51969). Ultimately, 103 articles were included in the qualitative synthesis, of which 58 are classified in ICC factors and 45 are classified in specific ethical issues of BCI research in six psychiatric disorders. Figure 1 below outlines the process of record selection. The BCI type distribution of including articles is shown in Figure 2.

Figure 1 PRISMA flow diagram outlining the process of study selection. The figure shows the number of records identified, screened, excluded, and included in the systematic review across PubMed, ScienceDirect, and Web of Science databases. ICC: Informed consent competency.

Figure 2 Distribution of brain-computer interface types in the studies included in the review. This figure illustrates the proportions of invasive, semi-invasive, and non-invasive brain-computer interface types among the selected studies.

Two standardized forms were constructed for data extraction according to ICC factors and ICC related ethical issues. Results of data narrative synthesis and tabular comparison were shown in Tables 1 and 2.

Table 1 Informed consent competency assessment in brain computer interface research involving six major psychiatric disorders.

Psychiatric disorders	BCI researches and applications	ICC challenges	ICC correlation and influence factors	Future research directions of ICC in BCI research
Schizophrenia	Non-invasive BCI: rt-fMRI neurofeedback for regulating abnormal neural connection[33]	Impaired DMC[27]	Symptoms severity correlation[27]	Expand clinical samples to include outpatients and patients with different symptom levels[27]
			Emotional processing and emotional regulation issues[28]
			Slower decision-making[27]
			Executive function impairment[27-29]
			Working memory deficit[27,29]
			Lower cognitive flexibility[27]	Explore neuroimaging integration in ICC[27]
			Auditory memory deficit[27]
			Rote memory[29]
			Visual memory deficit[29]
			Long-term memory impairment[27]
			Immediate memory impairment[27]	Include measures of motivation or effort to check for possible confounding effects[27]
			Processing speed deficit[27]
			Deficit in verbal memory and learning, semantic-categorical and phonological verbal fluency[27,29]
			Dysfunction of the orbitofrontal cortex (OFC) leads to reduced levels of ambiguous aversion[30-32]
Mood Disorders	Non-invasive BCI: aBCI using resting-state EEG for depression detection[43]	Influenced speed of decision-making but comparable advantageous choices[27]	Symptoms severity correlation[27]	Schizophrenia 1, 2, and 3 also applies to Mood Disorders[27]
			Slower decision-making[27]
			Long-term memory impairment[27]
			Immediate memory impairment[27]	Target deep brain regions using non-invasive brain stimulation methods to expand the applicability of deep brain stimulation, reduce risks, and maximize treatment effectiveness[41]
			Episodic memory impairment[35]
			Executive function deficits[27,35-38]
			Processing speed deficit[27,35]
			Verbal and non-verbal learning and memory[27]
			Variability in treatment response[27]	Developing an Emotional Blend BCI to support patients' brain activity and reduces addictive behaviors[42]
			Long-term remission maintenance[27]
			Emotional processing and emotional regulation issues[39,40]
Anorexia Nervosa	Non-invasive BCI: Rt-fMRI-BCI for self-regulation of anterior insular cortex activity[55]	Impaired DMC: Inability to make advantageous decisions, neglect of long-term consequences[46,47,51,52]	Body mass index is associated with decision-making ability[46]	Developing Personalized Intelligent Systems with BCI and Sensor Technologies[53,54]
			Impaired control system[48]	Developing mobile health technologies, digital twins, and social bots to provide real-time, personalized support for decision-making, behavior change, and daily living[53]
			Lack of feedback sensitivity[46-50]
			Overreaction to rewards associated with eating disorder behaviors[51]	Designing non-invasive brain stimulation methods to target deep brain regions can expand the use of deep brain stimulation therapy and reduce treatment risks[54]
			Anxiety and emotional dysfunction[53]
Alzheimer’s disease	Non-invasive BCI Invasive BCI[64,65]	Severely impaired DMC[62]	Executive function impaired[57,58]	Memory and Organizational Aids for Informed Consent[63]
			Short-term memory impairment[59]
			Episodic memory impairment[60,61]
			Working memory impairment[60,61]	Include neuropsychological tests to consent capacity research[62]
			Semantic knowledge impairment[59]
			Reasoning impairment[59]
			Verbal knowledge impairment[60,61]	Incorporate decision-making instrument instead of cognitive tests into clinical care[62]
			Processing speed impairment[60,61]
			Progressive cognitive decline[56,62]
			Influenced total score of the Alzheimer’s Disease Assessment Scale– Cognition (ADAS-Cog) and category fluency[57,58]
Parkinson disease	Invasive BCI[77]	Impaired DMC[67]	Dopamine deficiency impairs basal ganglia function including setting decision boundaries[67-69]	Whether dopamine itself affects using prior knowledge for decision making[67]
			Dysfunction in Basal Ganglia and Prefrontal Cortical Networks[70-72]
			Compensatory Shift to Drift Rate Adjustments[67,73]
			Impact of long-term dopamine replacement therapy[74-76]
Alcohol dependence	Non-invasive BCI: EEG-based BCI[88]	Impaired DMC: Characterized by increased risk-taking behavior and difficulty weighing long-term[79]	Executive functioning impairment[79-82]	Develop tasks that could generalize to the general spectrum of situations of decision-making under uncertainty[79]
			Working memory impairments (Storing and manipulating; Ospan Task)[79,83-87]
			Making disadvantageous decisions under ambiguity (Iowa Gambling Task)[79]	Design a task to consider interference from emotion and other cognitive processes and independently validate the effect of working memory executive function on risky decision making[79]
			Making more risky decisions under risk (Cups Task and Coin Flipping Task)[79]

BCI: Brain-computer interface; ICC: Informed consent competency; rt-fMRI: Real-time functional magnetic resonance imaging; aBCI: Affective brain-computer interface; EEG: Electroencephalogram; rTMS: Repetitive transcranial magnetic stimulation; tDCS: Transcranial direct-current stimulation; DBS: Deep brain stimulation; aDBS: Adaptive deep brain stimulation.

Table 2 Summary of specific ethical issues in brain computer interface researches and applications.

Key elements	Ref.	Specific ethical issues
Informed consent capacity	[90-97,126,130,136]	Patients with severe impairments (e.g. PD and AD) may lack capacity to consent, and psychiatric/ neurodegenerative patients with fluctuating decision-making abilities
		Requires assessing participants’ understanding of risks/benefits; surrogate informed consent (e.g., guardians) may be needed
		For pediatric BCI studies, legal guardians must consent, but child assent is critical for long-term engagement
		Safeguarding the rights and interests of vulnerable research subjects and ensuring their full understanding and voluntary participation
		Ensuring patient autonomy is essential and therefore emphasizes that the informed consent process should pay particular attention to the patient's expectation management and cognitive abilities
		BCI devices may affect cognitive abilities, e.g., STN may lead to ICD, which may affect the patient's ability to make decisions on his/her own
Ongoing consent & withdrawal	[94,98-103,114,135]	Irreversible interventions (e.g., invasive BCIs) require re-evaluating consent if identity/personality changes occur
		Informed consent should be a continuous process due to BCI’s long-term use
		Irreversible BCI implants may hinder withdrawal of consent (e.g., invasive BCIs)
		Refuse to complete the BCI research and remove BCI devices, due to their dependence on the technology and a lack of competency to fully understand future risks. This phenomenon is particularly common and distinct in the case of invasive BCIs, where withdrawal decisions are more complex and ethically challenging
Autonomy & control risks	[94,99,102-114,130]	Affective BCIs may manipulate emotions, threatening user autonomy
		Errors in decoding BCI signals could misinterpret intentions (e.g., legal declarations)
		BCIs may alter identity or personality, questioning consent validity post-implantation
		“Ambiguous agency” blurs user vs system control, undermining autonomy
Misunderstanding risks	[19,92,94,99,103,104,110,115-121,137]	Participants may lack understanding of BCI’s long-term effects or technical risks (e.g., surgery complications)
		Media hype (e.g., BCIs “read minds”) that distorts public understanding and consent validity and creates unrealistic expectations
Third-party consent ethics	[94,99,104,122-126,131,139]	Pediatric BCI decisions by guardians might conflict with a child’s evolving autonomy because they may not be able to undo the effects of these technologies in adulthood

BCI: Brain-computer interface; ICC: Informed consent competency; PD: Parkinson’s disease; AD: Alzheimer’s disease; STN: Stimulation of the thalamic nucleus; ICD: Impulse control disorder.

Existing major ICC assessment tools

Current methods for assessing ICC include a variety of vignette-based instruments and structured or semi-structured interviews designed to evaluate competency in specific contexts. Structured or semi-structured interviews focus on systematically assessing key cognitive functions such as understanding, appreciation, reasoning, and the ability to express a choice, allowing clinicians to evaluate participants' DMCs in a targeted and context-specific situation[14]. For example, the MacArthur Competence Assessment Tool for Clinical Research (MacCAT-CR) assesses DMC for research consent by evaluating understanding, appreciation, reasoning, and choice expression in a study-specific context, allowing flexible clinical judgment without absolute cutoff points[7]. The University of California Brief Assessment of Capacity to Consent (UBACC) is a 10-item screening tool that evaluates DMC in research participants by assessing their understanding and appreciation of study-related information, thereby identifying individuals who may require more thorough capacity assessments[15]. The 10 items cover the study's purpose, procedures, potential risks and benefits, voluntary nature, right to withdraw, alternatives to participation, confidentiality, study duration, and available contact information for further questions. The Hopemont Capacity Assessment Interview (HCAI) is a semi-structured tool designed to assess DMC, specifically targeting vulnerable populations like nursing home residents, by evaluating their ability to express a choice, understand and appreciate relevant information, and reason through decisions[16].

In summary, BCI clinical research is highly complex, requiring expertise from multiple disciplines. This complexity can make it challenging for participants, especially those without medical knowledge, to fully grasp the informed consent form[17]. Next, the influence of external factors makes informed consent in BCI research even more complicated. Healthcare providers have observed that patients often have unrealistically high expectations for interventions such as deep brain stimulation (DBS), which can stem from media portrayals of "miracle cures" and result in disappointment when the actual outcomes do not align with those expectations[18,19]. Moreover, BCI is mainly used in vulnerable populations with impairments in DMC, such as individuals with locked-in syndrome, schizophrenia, and bipolar disorders, which are often associated with significant DMC impairments[20]. Finally, as stated in the Declaration of Helsinki, if a potential research participant cannot provide fully informed consent but can give assent regarding their participation, the physician or qualified individual must seek this assent in addition to obtaining consent from the legally authorized representative or guardian. Any expressed preferences or values of the participant should be considered, and any dissent should be respected[21]. These factors together raise a significant concern: How to ensure BCI participants are truly informed and able to make autonomous decisions when external sources or mental diseases may influence their understanding.

We reviewed 103 studies to examine the methods used to assess ICC in BCI research and their applications for six major mental disorders. Our analysis found that aside from the MacArthur Competence Assessment Tool (MacCAT)[22,23], no other assessment tools have been applied in this context.

Assessment tools for ICC in BCI clinical research and application

Appelbaum and Grisso proposed the MacCAT-CR, containing four categories to determine ICC: Understanding, appreciating, reasoning, and expressing a choice[7]. Understanding is the capacity to grasp the information relevant to the decision, while appreciation refers to recognizing how this information applies to one’s own situation. Reasoning is the ability to evaluate the information to make a sound decision logically, and expressing a choice involves making and communicating a clear decision[7]. Similarly, Edelstein used these four categories in the HCAI to assess decision-making competence in elderly patients and populations with mental health issues. The HCAI employs constructed scenarios to evaluate a person’s capacity to absorb important knowledge (understanding), realize its personal consequences (appreciation), rationally weigh options (reasoning), and coherently explain a decision (expressing a choice)[16]. These areas constitute the basis for assessing ICC inside the instrument, ensuring a thorough assessment of an individual’s ability to make informed judgments. In contrast, the UBACC tool is a 10-item quick interview that only covers three of the four criteria: Understanding, appreciation, and reasoning[15]. However, while these tools are effective in general psychiatric clinical research settings, they have limitations when it comes to BCI-related clinical research. However, only MacCAT-CR has an application for mental illness-related BCI, while other tools do not. MacCAT helps assess DMC in BCI research by evaluating patients' understanding and reasoning abilities, particularly in cognitively impaired individuals. Still, its reliability is challenged by communication barriers and the accuracy of BCI-mediated responses in high-stakes decisions[22]. MacCAT-CR is also used in DBS research to assess DMC. However, its limitations include the persistence of therapeutic misconception, where patients may overestimate personal benefits or misunderstand research goals despite demonstrating high decisional capacity[24].

Existing tools for assessing DMC are primarily concentrated in DBS applications rather than broader BCI applications. Among these tools, MacCAT-CR is the only one currently used in mental illness-related BCI research, primarily for evaluating patients’ understanding and reasoning abilities. However, its application in BCI research faces several challenges, including communication barriers that affect reliability, the difficulty of accurately interpreting BCI-mediated responses in high-stakes decisions, and the persistence of therapeutic misconception, where patients may overestimate personal benefits or misinterpret research goals[22,24]. Also, research on its initial version, the MacCAT-T, found that factors such as aging, lack of self-awareness, and cognitive impairment were associated with impaired decision-making[25]. Additionally, informed consent in BCI clinical research presents unique ethical concerns, requiring enhanced protection for vulnerable participants, greater clarity in consent processes due to experimental risks and uncertainties, and strategies to overcome cognitive barriers that hinder participant understanding. The necessity of dynamic, informed consent is particularly critical in BCI trials, as DMC may fluctuate over time, necessitating ongoing assessments to ensure ethical and autonomous participation[17]. The ICC assessment frameworks for BCI participants’ ability to consent continue to raise questions. This systematic review aims to examine the ICC assessment in BCI clinical research involving psychiatric disorders.

ICC assessment in BCI research involving six major psychiatric disorders

We analyzed the influence or correlated factors of ICC in primary psychiatric disorders individually, which BCI research and applications focused on. We found that some mental disorders shared common factors, and specific factors were present in different disorders. Moreover, the major or mild neurocognitive disorder due to PD did not share common factors with other mental disorders, as shown in Table 1 and Figure 3.

Figure 3 Common and disorder-specific influencing or correlation factors affecting informed consent competency in six major psychiatric disorders. Executive function impairment, processing speed deficits, and memory dysfunctions are highlighted as common factors, while disorder-specific factors such as emotional dysregulation and dopamine-related decision boundary shifts are noted.

Table 1 provides a structured overview of ICC challenges, influencing and correlation factors, and future research directions for six major psychiatric disorders. Figure 3 shows ICC common and specific influencing or correlation factors in six major psychiatric disorders with ICC impairments. First, executive function impairment is a core shared feature across schizophrenia, mood disorders, AN, AD, and alcohol dependence. Executive function impairment is widely recognized as the most significant factor impacting ICC, as it undermines individuals’ ability to comprehend, deliberate, and make consistent decisions. Second, processing speed deficits are observed in schizophrenia, mood disorders, and AD, further complicating timely and efficient information processing during the consent process. Third, memory dysfunction-particularly episodic and working memory-contributes to compromised ICC. Working memory impairments are found in schizophrenia, AD, and alcohol dependence, while episodic memory impairments are evident in mood disorders and AD. Additional disorder-specific ICC correlation and influencing factors-for example, emotional dysregulation in AN and dopamine-related threshold shifts in PD-are summarized in Table 1 and Figure 3. These findings reinforce the common and different clinical features of disorder-specific ICC impairments in BCI research.

Schizophrenia

Schizophrenia is a chronic and complex mental disorder characterized by disturbances in thought, perception, behavior, and cognition, often including delusions, hallucinations, disorganized thinking, and cognitive impairments[26]. Individuals with schizophrenia may face many challenges in ICC due to a combination of symptom severity and cognitive-emotional impairments. Greater symptom severity has been associated with reduced decisional abilities and slower decision-making[27]. Difficulties in emotional processing and regulation further undermine the ability to rationally evaluate consent-related information[28]. Deficits across multiple clinical domains critically affect comprehension and reasoning. These include executive function impairment[27-29], working memory deficits[27,29], reduced cognitive flexibility[27], and auditory[27], visual[29], and rote memory impairments[29]. Additionally, immediate and long-term memory deficits[27], as well as slower processing speed[27], contribute to challenges in assimilating complex information. Difficulties in verbal memory, learning, and fluency-both semantic-categorical and phonological-also hinder effective communication and understanding[27,29]. Finally, orbitofrontal cortex dysfunction, which reduces aversion to ambiguous outcomes, may also impair decision-making in uncertain contexts[30-32]. Evidence indicates that BCI-based neurofeedback, particularly real-time functional magnetic resonance imaging (rt-fMRI), may assist in controlling abnormal neural connections in schizophrenia, providing a potential treatment option for individuals with brain network integration impairment[33].

Mood disorders

Mood disorders are a group of psychiatric illnesses that can simultaneously affect one’s emotions, energy, and motivation. The two most prominent examples are major depressive disorder and bipolar disorder[34]. Individuals with mood disorders face significant challenges to ICC, influenced by a range of cognitive and emotional impairments. These include deficits in executive function, and impairments in immediate, long-term, and episodic memory[27,35-37], decision-making speed[38]-all of which are closely associated with the severity of depressive symptoms[27,35-38]. Additional deficits in processing speed and both verbal and non-verbal learning and memory further undermine decisional abilities[27,35]. Emotional processing and regulation difficulties, as well as variability in treatment response and challenges in maintaining long-term remission, compound these impairments and may disrupt the capacity to understand, appreciate, and reason through research-related information[27,39-42]. Recent research has used affective BCI (aBCI) paired with resting-state electroencephalography (EEG) to detect depression and achieved high accuracy in distinguishing depressed teenagers from healthy controls[43]. In parallel, FDA-approved repetitive transcranial magnetic stimulation (rTMS), a non-invasive brain stimulation technique targeting the left prefrontal cortex, has been effectively applied as a novel treatment approach for depression[44]. These developments highlight the growing potential of non-invasive BCI technologies in diagnosing and treating mood disorders.

AN

AN is a severe psychiatric disorder characterized by restriction of energy intake or other behaviors for losing weight, like purging or over exercise, intense fear of gaining weight, and body scheme disorder[45]. Many studies have shown that patients with AN have impaired decision-making ability. In individuals with AN, ICC may be compromised due to impaired control system (executive function impairment), which includes sacrificed flexibility, goal-oriented behavior, and reduced feedback sensitivity. These limit the ability to evaluate outcomes and adjust behavior accordingly[46-50]. Decision-making ability is also associated with body mass index and may be further disrupted by an overreaction to rewards related to disordered eating behaviors[46,51,52]. Additionally, anxiety and emotional dysregulation present in this population can further impair the cognitive and emotional processes required for understanding and evaluating consent-related information[53,54]. Studies have shown that rt-fMRI neurofeedback can reduce hunger and is associated with the down-regulation of food cues, suggesting its potential for addressing maladaptive eating behaviors in AN[55].

Major or mild neurocognitive disorder due to AD

AD is a progressive neurodegenerative disorder and the leading cause of dementia, marked by cognitive decline and loss of independence in daily activities[56]. Individuals with AD demonstrate widespread deficits across multiple domains, including short-term, episodic, working, and semantic memory[57-59], as well as reduced verbal knowledge[59,60], processing speed[61], and executive functioning[59-61]. These impairments are associated with progressive cognitive decline and negatively impact performance on standard cognitive assessments such as the ADAS-Cog. The Cantonese version of the Mini-Mental State Examination and category verbal fluency tasks[56-58,62,63]. Liberati et al[64] indicated that BCI allows basic communication for AD patients in advanced stages by detecting cognitive and emotional states through non-invasive methods. Recent research has expanded on this by incorporating closed-loop BCIs with real-time neurofeedback and non-invasive brain stimulation techniques, such as TMS and transcranial direct-current stimulation, to enhance cognitive and affective functions in AD patients, with potential applications in rehabilitation and cognitive enhancement[65].

Major or mild neurocognitive disorder due to PD

PD is a progressive neurodegenerative movement disorder characterized by tremor, rigidity, bradykinesia, and postural instability[66]. ICC in PD is challenged by dopamine deficiency, which impairs basal ganglia function and disrupts the ability to set appropriate decision boundaries[67-69]. Dysfunction within basal ganglia and prefrontal cortical networks further compromises cognitive control and value-based decision-making[70-72]. Although patients may attempt to compensate by adjusting drift rates during decision-making[67,73], long-term dopamine replacement therapy may alter neural mechanisms and affect executive functioning, thereby influencing decisional capacity[74-76]. BCI applications for PD, particularly in the form of adaptive DBS, enhance neuromodulation by recording and processing neural signals in real-time, allowing for more precise and responsive symptom management compared to conventional DBS[77].

Alcohol dependence

Alcohol dependence disorder is a chronic condition characterized by excessive alcohol consumption leading to neuroadaptive changes in the brain, resulting in physical dependence, withdrawal symptoms upon cessation, and an increased vulnerability to relapse, often driven by heightened sensitivity to stress and alcohol-related cues[78]. Individuals with alcohol dependence often face significant challenges to ICC due to impairments in decision-making under both ambiguity and risk, as demonstrated by more risky and disadvantageous choices on tasks like the Iowa Gambling Task and the Cups Task[79]. These difficulties are compounded by deficits in executive functioning and working memory[79-81], particularly in storing and manipulating information essential for understanding and evaluating research participation[82-87]. EEG-based BCIs can detect alcohol cravings, monitor cognitive states, and assess neural changes associated with alcohol use disorders. Additionally, BCI technology holds promise in evaluating the severity of alcohol dependence and potentially predicting relapse risks, making it a valuable tool for clinical applications in addiction treatment[88].

In addition, co-morbidities exacerbate the severity of the disease, and their neuropsychological deficits may converge to create more severe impairments, with poorer long-term recovery and more severe psychosocial disabilities[36,89]. These findings suggest that BCI could play a role in early diagnosis and intervention for psychiatric disorders. These advancements, however, introduce new challenges in ensuring that patients fully understand the implications of BCI interventions. The complexity and experimental nature of BCI make it more difficult to obtain valid informed consent, emphasizing the need for rigorous ICC assessment in psychiatric populations.

Ethical challenges and specific aspects of ICC in BCI clinical research on psychiatric disorders

In the assessment of ICC in BCI research and applications for psychiatric disorders, it is important to consider not only the medical aspects but also the ethical challenges and disorder-specific concerns. Across the reviewed literature, several key themes emerged, highlighting both the potential benefits and ethical risks associated with BCI technology. The ethical considerations surrounding BCIs include ICC, ongoing consent and withdrawal, autonomy risks, misunderstanding of risks, and third-party consent ethics. Specific ethical issues are summarized in Table 2.

Table 2 summarizes the five core ICC related ethical issues in BCI research. First, regarding the ICC, patients with mental disorders and neurodegenerative diseases often do not have full capacity to consent due to severe impairments and need to resort to surrogate informed consent process. Persistent consent and withdrawal issues, on the other hand, are prominent in invasive BCIs, where patients may have difficulty autonomously deciding whether to terminate their participation in the study when technological dependence or changes in status occur. Autonomy risks include the potential for emotional BCI to manipulate the user's emotions, signal decoding errors that may misinterpret the user's intentions, and unclear boundaries of responsibility due to “fuzzy agency”. The risk of misunderstanding section warns that the public and patients do not have adequate knowledge of the long-term effects of BCI technology, its rationale, and privacy risks, and that unrealistic expectations are exacerbated by media hyperbole. The issue of third-party consent is particularly salient in pediatric research, where decisions made by legal guardians may conflict with the evolving autonomy of minors, making it difficult to undo the effects of the technology in the future. These ethical challenges require that BCI research should be designed and conducted with attention to fluctuations in participants' autonomy, protection of their rights, and the dynamics of the informed process.

Eleven papers in the reviewed literature emphasized the challenges of obtaining valid informed consent, particularly for patients with neurodegenerative diseases, psychiatric disorders, or fluctuating decision-making abilities[90-94]. Some patients may lack the capacity to fully understand the risks and benefits of BCIs, necessitating surrogate informed consent mechanisms, particularly in pediatric trials, where both parental permission and child assent are essential for long-term engagement[95,96]. Additionally, brain-to-brain interfaces introduce new ethical concerns, as direct brain communication complicates traditional models of individual agency and consent[97].

Ongoing consent

Consent in BCI research should not be a one-time event but a continuous process, particularly in the context of long-term or irreversible interventions such as DBS and invasive BCIs[94,98,99]. Patients who develop a strong dependency on their BCI devices may lack the competency to assess the risks of continued use, making voluntary withdrawal decisions ethically complex[22,100,101].

Autonomy and control risks

Autonomy concerns are a central theme in BCI ethics, particularly regarding the potential for aBCIs to manipulate emotions, intentions, and cognitive processes, which may compromise user autonomy[102-112]. Errors in decoding neural signals could lead to misinterpretation of a user’s intentions, raising concerns in legal and decision-making contexts[99,113]. Additionally, "ambiguous agency"-where control is shared between the user and the BCI-blurs responsibility, further complicating autonomy[104]. Emerging Web3-based decentralized BCIs present new challenges to identity and sovereignty, as cognitive functions may become distributed across networked systems, disrupting traditional notions of selfhood and control[106].

Misunderstanding and over expectation

Misconceptions about BCI technology are widespread, with media exaggerations often portraying BCIs as "mind-reading" or "thought-controlling" tools, which distorts public perception and creates unrealistic expectations[92,114-117]. Many patients and the general public misunderstand the long-term risks associated with BCI use, including surgical complications, data privacy breaches, and unintended neural changes[103,107,111,118,119]. Additionally, the "black-box problem" in AI-driven neurotechnology reduces transparency, making it difficult for users to understand how their thoughts and decisions are being processed and interpreted by the system[118-121].

Third party consent issues

The involvement of third parties in BCI decision-making raises significant ethical concerns, particularly in pediatric research, where legal guardians provide consent on behalf of children, potentially overlooking the child's evolving autonomy[122,123-125]. While legal guardians must authorize participation in invasive BCI research, gaining the child’s assent is crucial for ensuring long-term engagement and ethical integrity[126]. However, children may struggle to fully grasp the technical and ethical implications of their involvement, raising concerns about their ability to provide meaningful assent. Researchers must, therefore, carefully assess capacity, comprehension, and voluntariness to uphold ethical standards in pediatric BCI studies.

“Free and informed consent”

The 2024 version of the Helsinki Declaration emphasizes a paradigm shift toward a research participant-centered approach in medical research, positioning participants as collaborative partners with researchers. Notably, the traditional concept of "informed consent" has been replaced by "free and informed consent", thereby imbuing the capacity for informed consent with renewed significance. Informed consent refers to the process by which healthcare providers clearly communicate a procedure’s potential risks, benefits, and alternatives to ensure that patients can make a knowledgeable, voluntary decision[127]. More than a procedural step, informed consent safeguards autonomy and protects participants, particularly in psychiatric research, where mental health conditions can encompass the capacity to “consent to or refuse medical treatment” or “to be discharged against medical advice”[128]. In psychiatric research, ICC assessment is crucial for ensuring that individuals with psychiatric disorders can ethically and autonomously consent to participation, particularly in complex and high-risk fields such as BCI clinical research. Therefore, numerous techniques have been proposed to assist practitioners in determining the competence to consent to clinical research[7,15,16].

Vulnerable populations protection and dynamic consent

BCI is a highly complex new technology that requires multiple disciplines of knowledge and has unknown risks. Difficulties in understanding the closed-loop adaptive BCI concept and data security risk have been observed in healthy individuals with medical backgrounds, let alone patients with psychiatric disorders. Wang et al[17] also indicated the need for enhanced protections for vulnerable participants, the complexity and uncertainty of consent due to experimental risks, cognitive barriers that hinder participant understanding, and the necessity of dynamic, informed consent to accommodate changes in DMC over time. Moreover, the judgment of a clinician experienced in assessing a patient's competency may be influenced by the risk/benefit analysis of the patient's decision-making situation[129].

Socioeconomic inequality and the "involuntary nature" of choice

Additionally, social and economic constraints may limit patients' choices, compelling them to adopt BCIs as the only accessible medical option rather than a fully autonomous decision. Such systemic inequities, including financial barriers and restricted healthcare access, influence individual decision-making and participants’ ICC[18,104]. As stated in the Declaration of Helsinki (2024), when a potential research participant is unable to provide fully informed consent but can provide assent, the physician or other qualified professional must seek this assent alongside obtaining consent from the legally authorized representative. The participants’ expressed preferences and values should be taken into account, and any indication of dissent must be respected[21].

DISCUSSION

Analysis of autonomy in clinical research and practice

Informed consent for BCI has shown difficulty at all three stages: Adequate risk and benefit information, accurate understanding of the informed consent, and autonomous decision-making before and after the implantation. Decision-making is especially hard when patients may not truly understand the information due to clinical impairment or technical complexity. Patients with PD and AD have high variability in cognitive functioning, which may affect their ability to understand the risks and benefits of BCI technology and the purpose of the trial[130]. Such patients are particularly vulnerable during the autonomous decision-making phase and make judgments influenced by clincal factors such as their symptoms[131]. In addition to consent capacity being influenced by neurological symptoms, the invasive procedure of BCI, like electrode implantation, neural signal decoding process, and long-term side effects such as altered identity or impaired psychological integrity, can be very complicated. The complex nature makes it hard even for the general patient to understand the principles of the technique. Therefore, the cognitive deficits of patients with psychiatric disorders will only make this process even harder[17].

Ethical key issues in BCI informed consent

Ethical concerns related to informed consent capacity, autonomy, misunderstanding, ongoing consent, and third-party decision-making, particularly in vulnerable populations, have been well documented. These concerns include the inability to withdraw due to technological dependence, blurred responsibility in user-system interaction, and public misconceptions exacerbated by media portrayal. However, several issues remain under-addressed by current tools. Existing ICC assessment tools often rely on static, one-time evaluations and do not capture fluctuations in ICC over time. BCIs may restore communication abilities in ways that later conflict with initial proxy consent, and long-term studies increase the likelihood of autonomy changes. These factors indicate the urgent need for dynamic and context-sensitive ICC assessment mechanisms that can evolve alongside the participant’s condition.

Analysis of existing tools and intervention strategies

Although standardized tools like MacCAT-CR and UBACC have been validated for evaluating informed consent in conventional psychiatric research[7,15], they fail to consider the special multimodal risks connected with BCIs, such as neuro-privacy breaches, and aBCIs may generate concerns about "loss of self-control". However, current tools do not fully assess these special psychological impacts of BCI. For both patients’ benefits and ethical issues, improving patients’ understanding of informed consent and their ICC in the future is necessary. Flory and Emanuel[132] explored interventions to improve research participants' understanding of informed consent for research, and the results showed that talking one-on-one with participants was the most effective way of improving participants’ understanding. Also, enhancing consent forms, extending discussions, and using test/feedback techniques are effective methods in improving participant understanding[133,134]. In 2017, Wang et al[25] developed a one-week training course aimed at increasing capacity for informed consent research. As a result, after one week of training, community-stabilized schizophrenics could improve their understanding and appreciation, however, this training effect was not sustained after one year. These results suggest that more intensive or regular training may be needed to maintain long-term levels of competence in people with schizophrenia, particularly in their ability to understand the nature and consequences of participating in research.

Recommendations for future ICC assessment design

Future research should focus on developing BCI-specific ICC assessment tools that expand content to include unique risks such as neural data security, identity changes, and technological dependency[114]. A dynamic evaluation framework should be implemented, incorporating multiple assessment points throughout different treatment stages to track cognitive fluctuations[131]. Furthermore, cross-cultural adaptations are crucial to ensure informed consent procedures are region-specific. This can be achieved using guidelines like the Ethical Norms for Next-Generation AI to establish culturally appropriate consent standards[114]. Additionally, a multi-stakeholder decision-making mechanism should be created, focusing on prioritizing a patient’s regained communication abilities and incorporating structured dispute resolution processes[135]. A triadic communication model involving medical professionals, family members, and patients should be enhanced through scenario-based training to minimize biases in proxy decision-making[136]. Ethical and legal frameworks must also evolve, with a risk-based grading system differentiating ICC assessment intensity based on BCI invasiveness[90] and clear data governance protocols to define neural data ownership, usage, and deletion policies, preventing neuro-privacy breaches[114].

Therefore, optimizing the informed consent process in BCI research is imperative. Key strategies to enhance the process include prioritizing plain language explanations over technical jargon, for example, simplifying complex terms such as "unintended neurological plasticity" to more accessible descriptions of brain adaptation[137]. Evidence from Aliwi et al[138] also suggests that immersive virtual reality (VR)/augmented reality (AR) technologies can enhance patients' comprehension of medical information through intuitive aural/visual interaction, thereby improving their DMC. For instance, in optimizing informed consent processes, VR enables patients to observe three-dimensional dynamic visualization of treatment plans, including surgical step simulations, which can reduce reliance on complex terminology. These technologies significantly elevate patients' objective understanding of treatment feasibility and risks, as evidenced by a randomized controlled trial showing higher comprehension accuracy in VR groups compared to traditional 2D methods[138]. Furthermore, researcher preparedness plays a fundamental role in consent quality. Investigators must receive specialized training to effectively communicate BCI-specific risks, including unclear attribution between human and machine actions and identity destabilization resulting from the BCI devices’ future usage[130]. Moreover, procedural adaptations are necessary to enhance the effectiveness of informed consent. Multi-phase consent sessions (≥ 2 iterations) enable gradual assimilation of information, particularly regarding psychosocial risks like stigma amplification in psychiatric cohorts[130,139,140].

Cultural and linguistic considerations

The literature search discovered language and cultural differences significantly impacting the ability to obtain informed consent in BCI research. Different cultural contexts influence how individuals understand and approach medical technologies, shaping human behavior, cognition, emotions, and users' subjective experiences with health, illness, or technology use[141]. This is particularly true in the understanding of risk, the distinction between research and treatment, and the concept of autonomy[142]. Moreover, cultural norms surrounding authority and decision-making may affect how consent is obtained, especially in collectivist societies where family or community approval may take precedence over individual autonomy. This can make it difficult to determine whether consent is truly voluntary and informed, as individuals may feel pressured to comply with family or societal expectations rather than fully understanding the implications of their participation[143-145]. In addition, language is crucial in any BCI environment. Every BCI user must understand and interpret written or verbal instructions to follow commands or task directives. Therefore, BCI users' language abilities and cultural differences can influence their ability to comprehend BCI technology, which affects whether they can fully understand and engage in the informed consent process[141]. Furthermore, language barriers and varying levels of health literacy between different cultural groups may hinder effective communication of risks and benefits, further complicating the informed consent process. As BCI technology expands globally, it is essential to consider these cultural differences and ensure that informed consent procedures are culturally sensitive, transparent, and adaptable to diverse sociocultural contexts[108].

Related legal provisions

In the legal domain, competence is usually determined through formal statements that assume the user has a stable and continuous DMC[146]. However, the use of BCI introduces uncertainty into legal determinations. As emphasized by Zaubler et al[146], while clinical assessments of competence may be contextual and nuanced, legal statements of competence tend to be based on binary judgments that often fail to account for dynamic cognitive states affected by BCI systems. BCI may generate outputs based on subliminal neural signals or adaptive algorithms, resulting in user actions that are inconsistent with the original intent[147]. This inconsistency challenges the framework of current laws. Therefore, ICC in BCI clinical research will need to be aligned with future legal declarations.

Immersive technologies for ICC comprehension support

Recent studies have shown that immersive technologies, such as VR/AR, hold significant promise in enhancing patients’ understanding of procedures and associated risks. These tools can substantially improve decision-making accuracy by providing intuitive visual and spatial information[138]. Additionally, multi-phase consent processes and plain-language explanations can accommodate individuals with fluctuating cognitive status or low health literacy, making the consent process more inclusive and comprehensible[137,140].

In summary, a Five-Dimension BCI-specific ICC framework outlining these guiding principles and practical considerations for ICC assessment in psychiatric research is provided, including clinical, ethical, sociocultural, legal, and procedural dimensions. Five core ethical issues in BCI research should be addressed in the ICC assessment: BCI specificity, vulnerability, autonomy, dynamic ICC, comprehensiveness, and uncertainty. More details were shown in Table 3.

Table 3 Guiding framework for informed consent competency assessment in brain-computer interface research involving psychiatric disorders.

Category	Clinical elements	Ethical issues	Legal provisions	Sociocultural factors	Process of informed consent
ICC framework	Understanding	BCI specificity	Legal consistency	Justice	Intellectualization
	Appreciation	Vulnerability	Legal declaration	Cross-culture adaptation	Multi-dimension
	Reasoning	Autonomy			Subsidiarity
	Expressing a choice	Dynamic ICC			Convenience
	Biomarker of ICC (neuropsychological tests, neuroimaging integration)	Comprehensiveness			Accuracy
		Uncertainty			Personalization
					Real-time
					Time sufficiency

BCI: Brain-computer interface; ICC: Informed consent competency.

CONCLUSION

This paper offers a comprehensive review focused specifically on ICC in BCI psychiatric research, and aims to provide a foundational framework for future study. It fills a notable gap by systematically examining ICC in BCI clinical research across six major psychiatric disorders and proposes a Five-Dimension BCI-specific ICC framework-clinical, ethical, sociocultural, legal, and procedural-to propose future research in this area. The assessment of ICC in BCI clinical research and applications involving individuals with psychiatric disorders presents multidimensional challenges, such as the complexity of BCI technology, clinical characteristics of psychiatric disorders, specific ethical concerns, and cultural variations. The review found that the technology's complexity and potential long-term mental and identity impacts further complicate the ICC. Moreover, different psychiatric and neurodegenerative disorders pose specific challenges for ICC evaluation due to their disorder-unique clinical impairments and common deficits. Although all these disorders can affect ICC to some extent, the nature and severity of brain or neural network impairment vary and need to be explored. Additionally, cross-cultural differences in autonomy, health literacy, and decision-making norms can affect how patients interpret risks and their willingness to consent, especially when using proxy informed consent in collectivist societies. Traditional ICC assessment methods may not adequately reflect the multidimensional impairments in ICC experienced by psychiatric patients. As BCI research develops, a dynamic and multifaceted assessment system should be implemented, taking into account fluctuations in clinical characteristics and decision-making autonomy, the unique risks of BCI, and patients' changing needs over time. To address these challenges, future efforts should focus not only on assessing ICC but also on actively enhancing participants’ consent capacity through continuous capacity-building initiatives. This includes developing innovative assessment tools, refining evaluation mechanisms to account for BCI-specific ICC demands, and fostering interdisciplinary collaboration between neuroscientists, ethicists, and clinicians. By creating a scientifically rigorous and ethically inclusive ICC framework, BCI research can ensure regulatory compliance while prioritizing patient rights and well-being.