Background

Research on Brain-Computer Interfaces (BCIs) has seen a rapid development in the last 10 years [1]. A BCI is defined as a device that measures activity of the brain or central nervous system and converts these signals into artificial output [2]. BCIs thus detect and process brain and nervous system activity in order to control and direct external devices, such as personal computers, robotic arms, wheelchairs, or to activate a person’s own muscles [3,4,5,6,7,8]. A large area of application is in the use of BCIs in combination with medical devices, in order to increase communication or motor control for persons with physical impairments. BCIs have also been implemented in neuro-rehabilitation to improve neurological conditions such as motor paralysis after stroke or spinal cord injury [9, 10], as well as epilepsy [11]. Research has demonstrated that there are many uses for BCIs, including medical uses such as the treatment of psychiatric conditions, as well as non-medical uses such as neuro-enhancement or in gaming products [12,13,14].

Relevant brain activity from BCI users can be detected either with non-invasive (mostly with electroencephalography (EEG)) or invasive methods [6, 9, 15,16,17]. Three types of brain activity production and use by the technology can be distinguished [18]. (1) Passive BCIs use brain activity not voluntarily produced by the person, such as mental workload or affective states [18]. (2) Reactive BCIs are based on changes in brain activity which occur as a result of an individual’s voluntarily focused attention on a specific external stimulus (mostly visual, but also auditory or somatosensory) among a multitude of irrelevant stimuli [5, 18]. (3) Active BCIs require that the user applies a certain mental strategy (e.g. imagining moving a limb) [5, 18]. In virtually all BCI types, users receive real-time feedback on their brain activity output, through visual (most common), auditory, tactile, vestibular, or proprioceptive feedback [5]. Certain closed-loop applications of BCIs also apply stimulation to the brain by using electrodes which have been implanted in the brain, for instance in the treatment of epilepsy, or psychiatric disorders [19].

To understand BCI beyond its technical components and medical applications, significant social research is needed to grasp BCI in its practical and human dimension. Insights into the use of BCI and its impact on the user are necessary to develop the relevant knowledge and tools for ethical and legal evaluation.

The manner in which subjectivity or identity is molded by technology is an important issue for many different human-machine interactions, but especially in BCI applications (e.g. [20,21,22]). Entirely new characteristics involved in BCI, for instance incorporated technologies such as implants, or self-reliant and environment co-creating technologies, have the potential to influence the future of human-machine interaction. BCIs often employ these breaking threshold technologies and therefore present a crucial area of research in the molding of subjectivity through mediation with technology, and in human-machine interactions. Thus, it is clearly an essential matter to understand the first person perspective of BCI users, as well as the outlook of potential users and other BCI stakeholders.

In the interest of gaining an overview of existing studies, we have conducted a scoping review of published studies related to BCI that make use of social research methods. This scoping review is “aimed at mapping key concepts, types of evidence, and gaps in research related to” the field of BCI research “by systematically searching, selecting and synthesizing existing knowledge” [23]. The following paper should serve as a basis for future empirical studies and inform conceptual and ethical deliberations. We include studies considering various BCI stakeholders as well as quantitative and qualitative methods of social research.

Methods

A comprehensive review of the literature regarding brain-computer-interfaces and social research methods was conducted by following the five stages outlined by Arksey and O’Malley [24]. According to Levac et al. [25], we integrated further aspects such as a tabularly summaries and qualitative thematic analysis.

We aim to identify and characterize key social aspects of BCI use. In particular, the intent of the paper is to retrieve and synthesize existing scientific data on the perspective of BCI users. The focus of the review is not on technological and medical issues, but on psycho-social, personal, and ethical aspects of BCI use. For that purpose we turn to studies that employ social research methods as these methods can be regarded as a standard for reliable research outcomes with respect to societal and psycho-social practices. Hence, we selected the method of a scoping review with the intent of exploring the extent of research on the topic, summarizing findings and identifying research gaps. Given that the body of literature available is quite extensive and heterogeneous, a scoping review of the social and human implications of BCI use helps create a comprehensible overview for future research in this area of emerging and urgent relevance. This means, at the same time, that the body of literature is “not amenable to a more precise systematic review” [26].

Sources

Six databases were systematically searched and analyzed, namely PubMed, EMBASE, PSYCINFO, PSYNDEX, Sowiport, and SocINDEX due to the wide coverage offered in BCI-oriented research (across fields of medicine, psychology, and social sciences.) From March 1st to May 3rd 2017, two researchers performed the search. Publications dated later than May 3rd 2017 were not included in the systematic research, but may be referred to in the discussion section of the paper. All existing articles which met our research criteria were screened. In light of the novelty of BCI technology we expected further forms of publications and thus identified additional literature through searches in local university libraries, other internet sources, and by cross-referencing. We have drafted an internal review protocol which was not registered.

Inclusion/Exclusion criteria

Following a research consultation, we agreed on the following rationale for searching the databases. Two primary search terms and seven cross-references were used:

  • Search terms:

    • ○brain-computer* OR brain-machine*

  • Cross-references with:

    • ○ case study OR empirical OR interview OR qualitative OR quantitative OR questionnaire OR survey.

The cross-references were used to identify research on BCI that utilise social research methods. The search was performed on “all search terms” or “MeSH” terms (Pubmed).

“Methods of social research are the technical practices used to identify research questions, collect and analyze data, and present findings” [27]. Social research methods can be differentiated into quantitative and qualitative methods. The latter aim at the interpretation of meanings people assign to their actions, the former seeks to discover regularities in social phenomena. Quantitative methods are associated with surveys, experimental methods and the use of statistics. Qualitative methods often make use of interviews, focus groups, case studies or ethnographies, as well as methods of coding, theme, discourse or narrative analysis [27].

We limited the search in this case to the most relevant terms for the review. General key terms that cover the broad spectrum of research methods (empirical, quantitative, qualitative) were combined with specific terms to prioritize data gathering methods (i.e. case studies, interviews, questionnaires, and surveys).

The search was limited to literature published in English or German, two major scientific languages spoken by researchers in the field. No restriction was applied regarding the date of publication up to the date of the search.

To illustrate this procedure, one electronic search strategy conducted via PSYCINFO we will describe here in detail. The primary search term ‘brain-computer*’ was combined with the cross-reference ‘empirical’ using the AND-function. No specific field was selected. Furthermore, the search options were adjusted by choosing the search mode ‘Find all my search terms’ and selecting the publication type ‘All’. Initiating the search resulted in 25 hits. This procedure was repeated for each search term (brain-computer* and brain-machine*) and each of the seven cross-references separately. In total the research in PSYCINFO yielded 188 hits.

Analysis

Figure 1 provides an overview of the review procedure according to the PRISMA flow diagram [28]. The initial database research yielded 510 records. In addition, further publications found online or via a snow-balling method were manually added (n = 11) at the first stage of the research process. After removing duplicates and applying the inclusion/exclusion criteria, we identified 73 relevant publications. All of these publications were retrieved as full texts and analyzed using thematic analysis. The results were summarized in an Excel file. Each selected publication has been summarized and listed in tabular form (see Additional file 1).Footnote 1 We assessed risk of bias through evaluating and discussing our findings within a team of researchers at regular intervals.

Fig. 1
figure 1

The review process

The studies were categorized according to subgroups, reflecting the different groups of study participants (potential users, actual users, professionals, caregivers). Due to the amount of studies with potential and actual users, these groups each were divided further into two subgroups according to the methods used in the studies (namely quantitative and qualitative methods). This process allowed for a comprehensible display of the results and a more concise comparison within and among the various groups. In a second step, themes were identified by means of inductive reasoning in the style of the Qualitative Research Synthesis approach from Claire H. Major and Maggi Savin-Baden [29]. The themes are displayed either as columns in the tables or as bullet points.

Results

The final synthesis of the search undertaken contains a total of 73 publications of which 45 report quantitative studies, and ten refer to qualitative studies. Eighteen studies used a mixed-methods approach encompassing both quantitative and qualitative methods. Of the 73 publications, 13 report on the perspectives of potential users, i.e. persons who may have had a particular interest in using BCIs, but had no actual experience with BCIs. Fifty-eight publications focus on persons who have experienced BCIs either before or during the very study of the respective publication. Five studies additionally encompass caregivers/relatives. Seven studies address experts/professionals.

As outlined above, we obtained four groups of user studies (categorized along the axes – quantitative/qualitative studies and potential/actual users): quantitative studies with potential users (n = 11), qualitative studies with potential users (n = 2), quantitative studies with actual users (n = 48), and qualitative studies with actual users (n = 21). Studies employing mixed methods fall into both the quantitative and qualitative categories. Because of differences in terms of expectations and the impact of BCI among the users, the last category (qualitative studies with actual users) requires further differentiation. Accordingly, results of non-impaired users and users with physical impairments were assessed separately. The latter often are referred to in the BCI literature as “end-users”. (This terminology, however, could be quite confusing as not all BCI applications e.g. in the case of gaming, aim for persons with physical impairments as their end-users.)

Seventy-one of the 73 publications are focused on (potential or actual) BCI users. Two studies address BCI experts only. However, many of these 71 studies also encompass caregivers, experts, health care professionals or other BCI stakeholders.

Potential users

Research studies which have been conducted with potential users of BCIs mainly focus on assessing expectations that this particular group has about BCI as a so-called “assistive technology” device. These studies thus mostly limit these expectations to the question of the usability of the technology. The objective of such studies is to obtain information about the profile of the end-users for whom BCIs are being developed. The intent implicit therein is also to draw the developers’ attention to aspects deemed important by potential users, in order to meet user demand and adapt the technology accordingly.

Quantitative studies

There are 11 studies included in the review which assess the opinion of potential BCI users by means of quantitative methods (surveys/questionnaires) (Table 1). Three of the included quantitative studies evaluate non-impaired participants. They assess preferences of gamers [12], test for control aspects in gaming [30], and work on selection strategies for BCI use [31]. Eight studies in contrast focused on the needs of persons with physical impairments. The impairment- or medically-focused studies evaluated the interest in BCIs among potential users, marking their preferences regarding different features and functions, and requirements the technology is supposed to meet.

Table 1 Studies with potential users employing quantitative research methods

Qualitative studies

Only two studies assessed the opinions of potential users by means of qualitative methods. Klein et al. [32] set up a focus group and conducted telephone interviews with a total of 15 persons with implanted (open-loop) deep brain stimulation devices. The idea of BCI controlled closed-loop brain stimulation is met with acceptance among some potential users, while others feel ambivalent or opposed to the concept in principle. Comparing this hypothetical setting with the experiences of the study participants with open-loop brain stimulation, some participants welcomed the prospect of a self-controlled brain stimulation device while other participants maintained reservations. Some study participants expected an improved level of self-expression while others feared a distortion of their sense of self. The new brain stimulation device might also further restrain the user’s sense of accountability, in the case that others would hold the device responsible for their behavior or expressions. Furthermore, this application would require more trust in researchers to keep their data secure and confidential. This last point concerns the issue of meaningful consent. Participants mentioned difficulty in understanding all risks and implications of a closed-loop device both for themselves and for individuals with cognitive impairments.

Schicktanz, Amelung and Rieger [33] conducted open interviews with ten persons with various chronic conditions (e.g. amyotrophic lateral sclerosis (ALS), muscle atrophy, para−/tetraplegia), to whom they presented a video about a motor-readout BCI that controls a robotic arm. Participants hoped for more independence, self-control, privacy/intimacy, and better communication. Some were concerned about data protection and abuse. Additional concerns raised include the creation of self-transcending human-machine hybrids, as well as the fear for further dependencies, as the BCI necessitates service support and technological maintenance.

Actual users

For this review, 58 studies focused on actual or real life BCI users and their experiences with the technology, usually stemming from experimental BCI studies that tested various BCI models in terms of their functionality and feasibility. The feedback from the study participants was largely limited to usability aspects. In most studies, the objective was to assess the opinion of the participants regarding the tested model and to gain information about what improvements need to be made.

Quantitative studies

Forty-eight studies evaluated the users' opinions or assessed various personal factors of BCI users by employing quantitative methods (mostly in the form of questionnaires). Twenty studies employed BCIs in non-impaired  individuals, 17 studies tested BCIs in individuals with various physical impairments, and ten studies included participants from both groups. Most of these studies were concerned with the usability and feasibility of the technology. The performance of BCIs for users was assessed in aspects such as efficiency, effectiveness, and satisfaction. Furthermore, popular aspects assessed included cognitive or psychological factors, such as memory, attention, fatigue, mood, motivation, depression, and quality of life. Frequently applied tests were the NASA Task Load Index (NASA-TLX) for assessing subjective workload, the Quebec User Evaluation of Satisfaction with assistive Technology, Version 2.0 (QUEST 2.0) for measuring satisfaction, the Schedule for the Evaluation of Individual Quality of Life (SeiQol) for assessing the subjective quality of life, and the Questionnaire for Current Motivation (QCM) for testing the level of motivation. Performance was measured by the accuracy of the tasks, by the number of successful trials per session, or by means of the information transfer rate (ITR). Satisfaction was assessed via QUEST 2.0, visual analog scales (VAS) or the Assistive Technology Device Predisposition Assessment (ATD-PA) scale. Thematic aspects that were covered by the quantitative studies with actual users included:

These aspects were assessed for non-impaired participants as well as for participants with physical impairments. Quality of life was measured among the latter only.

Qualitative studies

Altogether from the qualitative studies evaluating BCI users’ perspectives, 21 studies emerged. Most of the qualitative studies used interview studies or focus groups. Several also opted for observations [81, 82] or discourse analysis [33]. One study focused on non-impaired participants only [54], 16 studies on participants with physical impairments, and four studies on a mixed population. In the breakdown of mixed methods studies, there are five studies including non-impaired participants (Table 2) and 20 studies with participants with physical impairments (Table 3). The objective of such qualitative studies was to improve and develop the technology of BCI as well as to embrace a medico-technological approach. Hence, qualitative data gaining methods were used for this purpose only. The studies which did not specify any qualitative data analyzing methods mostly ran statistical analyses and addressed the qualitative data as secondary [10, 36, 38, 40, 46, 49, 54, 56, 61, 62, 71, 72, 81, 83, 84]. Certain studies aimed also to address ethical aspects related to BCIs [83, 85,86,87]. Şahinol [82] researched BCIs as part of a genuine sociological study including elaborated sections about the methods employed in the technology’s use. Her study describes the mutual adaptation process between human and machine, resulting in the figure of a techno-cerebral subject.

Table 2 Studies with non-impaired participants employing qualitative research methods
Table 3 Studies with participants with physical impairments employing qualitative research methods

Themes that were assessed included opinions (judgements and attitudes) towards BCIs, or issues that arose during BCI testing. Others related to requests from the technology itself, such as aspects regarding social relations, quality of life, personality, and future BCI use.

Unlike non-impaired participants, persons with physical impairments have certain expectations of BCIs. Those affected by physical impairment tend to hope for more independence and social participation and expect from BCI use an increased quality of life. Some studies recognized the potential of BCI use to contribute to the user’s self-esteem and self-expression. BCIs are reported as bringing satisfaction and enjoyment, although space for improvement certainly remains in this area.

Other BCI stakeholders

Aside from the users themselves, BCI studies also often feature inquiries into the experience of relatives, caregivers, assistants, BCI professionals, developers, health care professionals or company representatives. Expert studies consisting of BCI professionals, game developers, and therapists/health care professionals are either focused on the potential and ethical aspects of BCI (BCI professionals) or on requests and concerns regarding BCI (therapists/health care professionals) (Table 4). These external stakeholder opinions are often addressed only marginally and sometimes are not displayed separately from the users’ perspective. A recurring group of stakeholders are caregivers (and/or relatives/assistants). Their participation forms part of focus groups or compliment the user’s perspective (Table 5). Their opinions hardly are presented separately from other results.

Table 4 Studies with experts/professionals
Table 5 Studies with caregivers/relatives

Discussion

The systematic search presented in this review offers an outline of BCI research which employs social research methods (in all of its methodological and thematic variety.) The occurrence of empty boxes in the tabular displays above indicates the heterogeneity of the BCI studies found. Some studies focus upon and include specific aspects studied, while others concentrate on divergent aspects.

Prominent topics within BCI research have been highlighted and research gaps identified. The majority of the studies addressed are concerned with the usability and feasibility of BCIs. This provides valuable and necessary knowledge for BCI professionals to improve and optimize use of the technology. Potential users in the studies examined point out requirements they would expect of BCIs, such as standards of speed, efficiency, and ease of use. BCI use should improve their overall situation and not place an extra burden on relatives or caregivers. The general level of satisfaction that BCIs experience among actual users proves that current technological development is likely on the right track. Actual users are able to pinpoint more specific aspects which may inspire further innovation. Besides the several points addressed by potential users, there was additional mention of a more appealing design of EEG caps, a preference for dry electrodes, the integration of particular devices (e.g. TV, phone), and a stress placed on the importance of home-based use.

Apart from the conceptual disparities between the different studies mentioned above, further difficulties arise when considering the huge variety of participant groups, not only between non-impaired and impaired individuals, but also with regard to the multitude of conditions seen among the participants with physical impairments. As the majority of studies contained a mixed sample of participants with various physical impairments, specific conclusions hardly can be drawn for BCI use for particular conditions. Only a handful of studies examined persons with a singular type of impairment (often in case studies with a single participant). These impairments included stroke [57, 63], cerebral palsy [67, 81], and essential tremor [84]. The exception thus is a total of 12 studies solely (or in combination with non-impaired participants) conducted with persons having ALS [37, 39,40,41, 50, 58, 59, 74, 88,89,90,91]. These results do not differ significantly from the other studies. One difference might be that persons with ALS (especially those in advanced stages) show a higher degree of interest in BCIs. This may be explained by the dismal outlook for mobility from this disease, which gradually robs patients of the ability to move their limbs, communicate verbally and interact with their environment, approaching a state of locked-in syndrome. In other conditions, patients may retain certain motor functions of their body that may allow them to use other technological devices, such as peripheral neuroprostheses. Few studies have been conducted with BCI professionals.

Very little interest has been placed so far on the experiential dimension of BCI users beyond usability aspects. Where the BCI literature examines “user experience” - sometimes simply called “UX” [47] - or pursues a user-centred approach, the respective studies apply psychological factors like mood, motivation, or depression, the participants’ quality of life, satisfaction, their opinions, judgements, and requests to BCI use. In contrast to the user’s experience with BCI, the user’s experience of BCI has been hardly researched. This outlook would comprise questions towards BCI use from the point of view of philosophy and social sciences, broadly describable as the What is it like? – perspective. What is it like to use a BCI? What is it like to act by using a BCI? And therefore: What is it like to act without using my body? What is it like to be hooked up to a machine or computer? Is it (still) me that is acting within this BCI system or is it some kind of human-machine hybrid? These are questions that address topics like agency, autonomy, responsibility, accountability, self-image, identity, hybridization, or artificial intelligence.

Dealing with these questions will be of major relevance, not only because the BCI relies on a new form of connection between human and machine, but also because the technology produces bodily experiences the user would not have otherwise. For example, performing an action without moving a body part opposes our common understanding and sensation of being an agent and can manifest itself as a new experience.

First attempts in this direction have been made by Grübler and Hildt [85,86,87]. Their research study raised the question of transparency, i.e. the ability to operate a tool without having to consciously apply its operating instructions. Applied to the example of BCIs, this would mean operating a BCI without focusing on a mental strategy. The study in question also asked participants whether they felt themselves to be a part of a functional unit with the computer. Some participants reported experiencing transparency, but fewer participants reported having felt being a part of a functional unit with the computer. The authors postulate a discrepancy between these two judgements.

Şahinol [82] points to several aspects related to the sense of agency and shared agency with the computer, revealing uncertainties among the users about how to make sense of BCI activities. The objective of Şahinol’s study, however, is about the adaptation process between humans and machines and therefore does not elaborate on the subjective user perspective regarding these aspects. Two studies published after the completion of our data collection and review, specifically by Gilbert et al., questioned users with implanted BCIs for the treatment of epilepsy. These studies yielded valuable insights regarding control and self-perception [92] as well as autonomy and deliberation processes [93]. This step would be crucial in working to close the research gap outlined above.

Among studies with BCI professionals, the focus lies on ethics. Crucial areas of inquiry like informed consent, managing users’ expectations, and psycho-social consequences of BCI use are explored. Another recent study, conducted by Specker Sullivan et al. with neural engineers [94], was directed towards an improved sensitization of the inclusion of users in the development process of BCIs. The user-centred design endorsed in the paper is present among various studies of this review’s body of literature [38, 57, 58, 62, 72, 91]. Therapists and health care professionals are skeptical about BCIs and require certain improvements of the technology. Caregivers’ perspectives resemble users’ opinions. A study focusing on caregivers only, however, has not been conducted so far.

As most of the studies discussed in this article are guided by a medico-technological approach, ethical issues are hardly being addressed and under-researched. This observation about the field of inquiry in BCIs is discussed in detail by Specker Sullivan and Illes [95]. When ethical aspects are addressed, it is in accordance with the ethical aspects addressed in the ethics literature, as shown by Burwell et al. [96]. Matters such as autonomy, agency, personality, safety or privacy are comprised.

Our own team of researchers encourages more empirical work on these matters. While philosophical in nature, matters such as agency, autonomy and responsibility are also highly relevant for legal and policy-making affairs. This particularly depends on the part the user plays within the loop of the BCI, and its data collection. Particular closed-loop neuro devices such as closed-loop DBS modulate stimulations outside the awareness of the user and therefore cause moral and legal issues of accountability [97]. In BCIs the user receives feedback in some form (visual, auditory or haptic). She usually is nominally aware of this feedback and is given the opportunity to react to it deliberatively. The user “stays in the loop” even at least partly when connected to some autonomous system and hence gives rise to a different situation of accountability [98]. To evaluate these aspects in practice would yield important insights to inform moral, legal and political debates.

Limitations

This study has several limitations. First of all, a scoping review cannot guarantee that all scientific literature is exhaustively found and analyzed. The literature output is confined to studies that explicitly study brain-computer- or brain-machine-interfaces or imply one of the terms as a key term. Some technologies that are labelled under “neurofeedback”, “closed-loop” or “predictive brain devices” may also qualify as brain-computer-interfaces. However, as not all of these are de facto brain-computer-interfaces and to keep the task at hand feasible, we confined our search to the search terms brain-computer* and brain-machine*. For the same reason we disregarded comparisons to other technologies such as open-loop deep brain stimulation which may render comparable results (e.g. [99]). Among the studies examined, we added manually some publications (books and book chapters) which are not peer-reviewed publications.

In order to manage the total search outcome of 73 publications, various differentiations have been neglected and need further examination: BCI varies widely in terms of its technical set-up, of which each application and model would require particular attention, e.g. in terms of measurement (invasive - non-invasive, EEG, NIRS, fMRT, ECoG, or others), mental strategy (selective attention, motor imagery), stimuli set-up (visual, auditory, haptic), application (communication program, prostheses, computer game, or others), or type of neurofeedback (displayed on monitor, successful movements, or others). Also comparisons between non-impaired participants and participants with physical impairments as well as between different impairments deserve more detailed attention. The scoping review method is a useful tool to map and synthesis large bodies of literature which comes at the expense of a detailed analysis of the results. The various themes identified in this review therefore deserve a further, elaborated examination.

Conclusion

A great deal of research has been conducted on the perspectives of potential and actual BCI users. These opinions, and emerging social research data, are key in advancing the development of user-appropriate, humane, and successful BCI technology. BCIs have much to offer: the ability to increase quality of life, enhance social life, and contribute to a higher level of self-determination and independence for persons with physical impairments. At the same time, this technology can lead to impingements on human autonomy, psychological frustration, the creation of dependency, and the causing of confusion regarding user self-perception. The qualitative self-experience of BCI users, i.e. aspects related to personal identity, agency, and responsibility, has hardly been examined thoroughly. This is due to the delicate and demanding nature of this research as well as a limited number of study participants until now. Nonetheless, as questions regarding the experience of BCI users are crucial for evaluating ethical and societal aspects of an emerging technology, more empirical research on these matters is deeply encouraged by researchers involved.