Identifying P300 Brain-Computer Interface Training Strategies for AAC in Children: A Focus Group Study

Abstract

The integration of Brain-Computer Interface (BCI) technology into Augmentative and Alternative Communication (AAC) systems introduces new complexities in training, particularly for children with diverse cognitive, sensory, motor, and linguistic abilities. Effective AAC training is crucial for enabling individuals to achieve personal goals and enhance social participation. This study aimed to explore potential training strategies for children using P300-based BCI-AAC systems through focus group discussions with experts in AAC and BCI technologies. Participants identified six key themes for effective training: (1) Scaffolding—developing adaptive systems tailored to each child’s developmental level, including pre-teaching, visual display adaptations, and gamification; (2) Verbal Instructions—emphasizing the use of clear, simple language and spoken prompts; (3) Feedback—incorporating immediate feedback and biofeedback methods to reinforce learning; (4) Positioning—ensuring proper trunk stability and addressing electrode placement; (5) Modeling and Physical Supports—using physical cues and demonstrating BCI-AAC use; and (6) Considerations for Visual Impairment—accommodating cortical visual impairment (CVI) with suitable stimuli and environmental adjustments. These insights provide an initial foundation for identifying BCI-AAC training strategies for P300-BCI-AAC use by children. Further systematic research involving end users and their support networks, along with other professions, is needed to validate, enhance, and identify new intervention approaches that support communication outcomes for children with varying needs.

Augmentative and alternative communication (AAC) encompasses a diverse range of strategies and tools designed to support the participation of individuals who cannot use their natural speech to meet their daily communication needs. AAC serves a broad spectrum of individuals, including those with conditions such as amyotrophic lateral sclerosis (ALS), cerebral palsy, autism spectrum disorder, and developmental disabilities. For instance, high-tech AAC solutions may include speech-generating devices equipped with software that allows users to select and vocalize words and phrases, while low-tech options could involve communication boards or books with symbols or pictures representing various concepts and messages. These tools cater to individuals with varying levels of communication abilities and aim to provide them with the means to interact and engage with others (Beukelman & Light, 2020).

Noninvasive brain-computer interface (BCI) is an emerging technology aiming to support AAC access, particularly for individuals with severe physical impairments. Among the diverse approaches to BCI for AAC access (BCI-AAC), the utilization of the P300 event-related potential (ERP or brain signal) has received significant attention (Brumberg et al., 2018; Donchin et al., 2000). Using a P300-based BCI-AAC commonly involves a person focusing their attention on a specific item displayed on a computer screen. These items might be letters, symbols, or pictures. While the person is concentrating on the item they want to select, the computer randomly highlights all the different items on the screen. When the desired item to which the person is attending is flashed, the P300 brain response may be triggered. P300-BCI-AAC algorithms then attempt to relate this P300 brain response with a particular flashing item and select that item as the person’s intended choice. In this way, a person may be able to communicate or control devices using only their brain activity, without needing to move any muscles (Pitt et al., 2019a).

While the majority of previous research on BCI-AAC has concentrated on adults, there is now a growing emphasis on including children. This shift acknowledges the distinctive needs and obstacles that children encounter in communication (e.g., Huggins et al., 2022; Kinney-Lang et al, 2020; Pitt et al., 2022, Pitt et al., 2024a). The extension of BCI research to encompass children presents new opportunities for improving communication access and engagement for individuals with conditions like cerebral palsy, who could potentially benefit from BCI-AAC technologies. However, this expansion also brings forth numerous challenges that must be addressed. These challenges include considerations such as the developing nature of the pediatric brain, the presence of muscular artifacts in EEG signals, the necessity for engaging interface designs, the diverse cognitive, sensory, motor, and linguistic profiles of individuals, and the importance of developing effective and engaging training protocols (Huggins et al., 2017; 2022; Pitt et al., 2019b; Pitt & Dietz, 2022).

Training and opportunities for practice are essential components of AAC intervention, helping individuals to achieve personal goals, strengthen relationships, and engage more fully in society (Blackstone et al., 2007). These interventions encompass a range of strategies tailored to meet each individual’s unique needs. For example, game-based tasks may be used to initially engage children using AAC in the learning process (Mandak et al., 2022; Pitt et al., 2019b). Additionally, widely implemented methods for enhancing AAC outcomes include drill-based approaches, explicit instruction, and opportunities for independent exploration (McNaughton et al., 2008; Rackensperger et al., 2005). These approaches are frequently complemented by scaffolding methods, which provide varying levels of support to assist in task completion (Light et al., 2008; Rackensperger et al., 2005). Such scaffolding techniques aim to gradually empower individuals to perform tasks independently while offering support as needed. Moreover, a variety of research exists supporting the use of modeling, also known as aided language modeling, and responsive partner instruction in enhancing AAC proficiency (Biggs et al., 2019; Sennott et al., 2016; Wandin et al., 2023). In AAC modeling, communication partners demonstrate accessing the communication display, such as selecting items through touch, while concurrently communicating verbally (Wandin et al., 2023).

The integration of BCI-AAC technology introduces new challenges in AAC training. These challenges are especially pronounced when addressing the diverse cognitive, sensory, motor, and linguistic needs of children. For instance, conventional AAC intervention methods, like modeling, rely heavily on physical actions (e.g., demonstrating AAC selection through touch). However, BCI-AAC devices are governed by brain activity, influenced by cognitive processes such as attention. This means providing models and verbal description of the P300-BCI-AAC control task may be challenging (Pitt et al., 2024b). Furthermore, adaptations to the visual P300-BCI-AAC interface can impact the targeted brain signal, thereby affecting BCI-AAC accuracy (Pitt et al., 2019a). To address the lack of P300-BCI-AAC training resources for children, initial studies have begun exploring potential approaches. For instance, a recent study by Huggins et al. (2022) discussed using videos to link the mental task required for P300-BCI-AAC control and response. They also explored varying verbal instructions, such as counting or using cues like ‘now’ when the item is flashed. However, research is still in its early stages, and identifying effective strategies for P300-BCI-AAC training is critical for advancing BCI-AAC success (Pitt et al., 2019b).

Implementation science can play a role in helping translate research findings into practical applications, bolstering the integration of technologies and interventions into clinical practice and daily routines (Douglas et al. 2022; Pitt & Dietz, 2022). Early-stage implementation efforts should consider stakeholder engagement, such as caregivers, clinicians, researchers, and those who use AAC, to inform the development and refinement of BCI-AAC systems. Involving stakeholders promotes a collaborative approach to BCI development and deployment, which can be enhanced by considering diverse perspectives. As communication professionals who play an integral role in prescribing AAC systems and supporting their use (Beukelman, & Light, 2020), speech-language pathologists (SLPs) may provide valuable perspectives on clinically focused P300-BCI-AAC development (Nijboer et al., 2014; Pitt et al., 2022). Therefore, based on the complexities and unique needs associated with P300 BCI-AAC use in children, and the principles of implementation science aimed at promoting the integration of research into practice, the aims of this investigation are to identify initial directions for P300 BCI-AAC training in children. P300-BCI-AAC was selected as the subject of inquiry due to its significant development focus (Peters et al., 2022). Through focus group dialogue, findings endeavor to guide future research directions and help the field of BCI-AAC consider inclusive and supportive training methods.

Methods

In our study, qualitative focus groups interviews were utilized to identify perspectives on strategies for BCI-AAC training. Participants were recruited using a combination of established researcher networks and word-of-mouth referrals to help provide a broad pool of participants, including those with practical experience in both BCI-AAC and AAC. Participants identified through researcher networks were contacted via email. Participants were not financially compensated for their participation in the focus group.

Two virtual focus groups were conducted based on guidelines by Guest et al., (2017). The focus groups followed a semi structured guide that was developed to address gaps in existing literature regarding P300-BCI-AAC training. The focus group question guide inquired about verbal instruction, visual supports, and modeling, alongside other training considerations, such as positioning and environment factors. The focus group question guide is provided in supplemental material A. Before commencing the interviews, participants were provided with a PowerPoint presentation. The presentation included descriptions of P300-BCI-AAC and a brief discussion of existing training methods. The presentation included links to publicly available videos demonstrating P300-BCI-AAC use. This presentation aimed to standardize participants’ knowledge and provide a foundation for informed discussion. To ensure participants’ comfort and understanding, the lead author reviewed the presentation at the beginning of the focus group, confirmed participant comprehension, and addressed any queries before proceeding with the focus groups. Focus groups were conducted via Zoom video conferencing, lasting approximately 60 minutes. All participants joined the online focus group using a link sent to their provided email addresses. They connected their cameras and microphones to give real-time comments and feedback during the sessions. Throughout the focus groups, additional inquiries and requests for clarification were made to ensure precise coding and deeper exploration of participants’ viewpoints. The study received approval from the Institutional Review Boards at the University of Nebraska-Lincoln and Nationwide Children’s Hospital (approval number: 00022837).

Participants

Focus group participants (8 females, 1 male), included eight SLPs, four based in Canada, and four in the United States, along with one BCI-AAC engineer (P7), based in the United States. One participant requested not to disclose their age. The age demographics of the other participants were, on aggregate, M = 45.6 years, SD = 11.7, range = 30–58. Recruitment of nine participants is consistent with prior published reports on focus group procedures in the field of AAC (Boster & McCarthy, 2018; Oommen et al., 2015). All participants had experience working with individuals who have severe speech and physical impairments. Participants employed in a range of settings were purposefully targeted to help ensure inclusion of a diverse array of experiences. Further, individuals with clinical backgrounds in working with those who have severe speech and physical impairments were targeted to help integrate BCI-AAC techniques with existing clinical practices. Only individuals with a minimum of two years’ experience in AAC and/or BCI-AAC were included in this study. All participants exceeded our minimum inclusion criteria for AAC expertise (M = 20.2 years, SD = 10.7, range = 7–36). See Table 1 for further demographic information.

Table 1.

Participant Demographics

Participant	Years of Experience in AAC	Setting	AAC Experience Related to Alternative Access for Children	Experience with BCI	Age Group Primarily Working With
P1	32	Outpatient Pediatric Clinic	Scanning and Eye-gaze	Read/heard about BCI	Children
P2	7	Academic Medical Center	Scanning and Eye-gaze	Assisted with BCI design, testing and training	Children and Adults
P3	8	Outpatient Pediatric Hospital	Scanning and Eye-gaze	Read/heard about BCI	Children
P4	26	Outpatient Hospital	Scanning and Limited Experience with Eye-gaze	Assisting with BCI research	Adults, but has had 10 years prior experience with children
P5	10	University Outpatient Clinic	Scanning and Eye-gaze	Read/heard about BCI	Children and adults
P6	27	School	Scanning and Eye-gaze	Tried a BCI and assisting in research	Children
P7	16	Academic Research	None	Leads BCI development, testing and implementation	Children and Adults
P8	36	University Outpatient Clinic/School	Scanning and Eye-gaze	Read/heard about BCI and assisting with research	Children
P9	20	Pediatric Hospital	Scanning and Eye-gaze	Assisting with BCI research	Children

Research Team

The research team consisted of three individuals. Two of them held certifications as SLPs. Both SLPs possessed knowledge in AAC implementation and research, alongside P300-BCI-AAC research and/or implementation. The third team member was an undergraduate student majoring in SLP who had experience with P300-BCI-AAC implementation and research.

Data Analysis

Data analysis procedures drew from established qualitative methodologies in AAC research (e.g., O’Neil & Wilkinson, 2022). Initially, both focus groups were recorded and transcribed verbatim, with a secondary graduate assistant verifying transcription accuracy through consensus discussions. Subsequently, the transcribed files were imported into NVivo software (QSR International, 2018) for systematic analysis.

Adopting a grounded theory approach (Gibbs, 2008), focus group themes were organized using NVivo’s coding features, employing a constant comparison method to integrate new data into the existing coding structure (Creswell, 2012). This iterative process led to the development of a comprehensive codebook, encompassing one major theme, six subthemes, and 18 example codes (refer to supplementary material B). There was overlap in discussed BCI-AAC training strategies across focus group questions. For instance, modeling techniques usually include additional verbal supports. Therefore, efforts were made to consolidate question responses into single representative codes to enhance data clarity and reliability.

Following initial coding of the data, the lead author and trained assistant conducted a thorough evaluation of codebook consistency by reevaluating all transcripts, resolving discrepancies through collaborative discussion to reach consensus. The multitude of themes identified in this study minimizes the likelihood of chance agreements in coding. Thus, percent agreement (Syed & Nelson, 2015) was employed to assess reliability. In more detail, reliability was evaluated by having an independent evaluator code a randomly selected focus group transcript (50% of the data), using the developed codebook. Intercoder reliability was independently conducted until a minimum of 80% accuracy (O’Neil & Wilkinson, 2022) was attained by a trained assistant. In our investigation, we surpassed this threshold, achieving an intercoder reliability of 100% for the subtheme and 90% at the example code level, for the selected focus group transcript.

Data credibility

Data credibility was upheld through various techniques, including member checking, and triangulation (e.g., Creswell, 2012; Gibbs, 2008). Specifically, the second author conducted peer review of the study methods, findings, and conclusions. Additionally, member-checking procedures were carried out during and after the focus group. During the focus group, summary statements were provided to ensure interviewer comprehension, and participants were prompted to clarify any ambiguous statements. Following each focus group, participants were sent a discussion summary for confirmation of accurate representation of their ideas. Eight out of nine participants responded to our request to review the discussion summary and indicated agreement. The other participant did not respond to the author’s request to review the discussion summary. To help mitigate potential biases, triangulation was achieved by employing a team approach during data analysis. This team approached included both authors and a trained research assistant.

Results

The subsequent section delineates focus group data structured according to theme, subthemes, and examples. Table 2 offers a concise overview, with the complete codebook available in supplementary material B.

Table 2.

Themes, Subthemes and Example Codes

Theme	Subtheme	Example
BCI-AAC Training	Verbal Instruction	Considering previous verbal AAC instruction and BCI as an addition to current AAC Explicit, concrete, and simple language with additional processing time Metaphors may be better for professionals as maybe too complex for kids Saying now and looking and counting prompts
	Scaffolding	Begin with attention to physical objects and pre teaching or natural activities General need to scaffold and build skills like attention Having control over selection accuracy [positive feedback] Need to consider flexible and supportive systems based on developmental level that provide opportunities for learning and communication Providing opportunities for communication and social interaction Visual display and adaptations gamification and simplifications
	Providing Feedback	Partner feedback Providing Biofeedback (alpha) Having the display item change in quality as the BCI-AAC moves toward selecting
	Positioning	Engineering solution for posterior electrode issues Stable base of support
	Modeling and Physical Supports	Pointing to the item Tapping shoulder Turn taking and pretending to use the BCI while talking aloud
	Considerations for CVI	Considering those with cortical or cerebral visual impairment (CVI)

BCI-AAC Training

Scaffolding

Scaffolding describes what steps can be taken for an individual with no or limited knowledge of BCI-AAC control to grow in P300-BCI-AAC proficiency. During the focus group, the importance of scaffolding to build foundational skills such as attention was noted. For instance, P4 described “…like a mock BCI that could just iteratively build over time sort of starts very simple.” In supporting scaffolding implementation, participants also highlighted the necessity of implementing flexible and supportive systems that may be adapted to match the developmental level and trajectory of each individual child, so the child is not expected to adapt to the BCI-AAC technology.

Regarding more specific scaffolding techniques, participants discussed the possible value of pre-teaching BCI-AAC concepts using real-world objects and natural activities for some children. For example, engaging children in familiar games or activities, such as having the individual find physical objects or promoting sustained focus on TV commercials, could help bridge the gap between tangible experiences and the more abstract nature of computer based BCIs. In this regard, P6 said, “I think you have to pre train this stuff off the computer more tangibly with objects and with social before… just because that computer is so abstracted”. Participants also discussed the benefit of allowing the BCI-AAC trainer to have some form of manual control over selection accuracy to speed up the rate of selection, reinforce desired behaviors, and boost the user’s confidence. For instance, the trainer could force the BCI-AAC to quickly make an accurate selection, even if the BCI-AAC algorithm was predicting an incorrect selection. The importance of providing opportunities for communication and social interaction was also raised by participants. Within this topic, focus group attendees described adapting BCI-AAC activities to align with children’s interests and motivations, considering a connection between BCI-AAC use and real-world communication, and maintaining a social component. Regarding a social component, P6 said,

“So, I’m always like, how do I keep myself in the game? If I’m doing something on technology, how do I keep myself in there so that there’s some social piece? And then what? How do I connect it with something real?”

Finally, an array of visual display adaptations, gamifications, and display simplifications were considered for promoting the initial acquisition of P300-BCI-AAC skills. These adaptions included: providing initial exposure to the likely unfamiliar flashing interface without the expectation of the individual trying to make an item selection to promote familiarity; creating a game out of the control task to lower language burdens (e.g., similar to Look to Learn games, where engaging in a P300-BCI-AAC-related task, such as extending periods of sustained attention, allows the player to actively participate in the game); drawing extra attention to the target item by making it a) something fun, interesting, and preferred (e.g., a picture of a favorite toy or close family member), b) move, c) change color, d) make fun and engaging sounds, and e) slowing down highlighting rate, possibly with a longer highlighting time of the target item or pause feature to provide time for reinforcement of task instructions. Further, participants discussed reducing the number of items in the visual array, along with possibly using a fly swatter with an area cut out in the center to draw attention to a target item. In more detail, placing an external object in front of the screen to block or highlight specific areas may aid in focusing attention. This approach is similar to how a reading guide strip helps individuals concentrate on a single line of text by obscuring the surrounding lines. However, it was noted that decreasing the number of items in the visual display also changes the visual statistics of the interface (i.e., there are now less non-target distractors), which may negatively impact P300 elicitation. Finally, the group noted using a physical symbol (e.g., printed on a piece of paper) to help reinforce the connection with the symbol on the computer display, along with simplifying the environment by limiting distractions.

Verbal Instruction

Focus group participants discussed verbal instruction, describing what can be said and how something can be said for users to understand how to use P300-BCI-AAC. They emphasized the importance of using explicit, concrete, and simple language along with additional cognitive processing time when instructing users on BCI-AAC usage. This approach involved breaking down tasks into short, simple phrases—sometimes just one or two words—and pausing between instructions to ensure comprehension. For example, P6 said, “So, I tend to just give things in a very simple, very limited, verbal, with lots of kind of pausing and processing time in between”. Participants also highlighted the efficacy of spoken prompts, such as the person providing the training saying “now” or counting, to indicate when users should focus on a specific target. In this regard, P2 discussed “And so we trained people by saying like now, or like when, when, this letter came on the screen, like we gave like a, a spoken prompt to them”.

Integrating BCI-AAC with previously learned AAC methods, like switch scanning or eye gaze systems, was another subtheme raised by participants. Participants discussed that if the person learning P300-BCI-AAC has previously learned another AAC method, the successful teaching instructions provided to teach this other AAC method may be leveraged to promote familiarity, streamline the BCI-AAC learning process, and reducing cognitive load and confusion. For instance, P2 stated:

“Somehow, if they’re using switch scanning or eye gaze, if you could train the task with that response like, if that was already something, you know, we’ve been working on switch scanning, and they know to choose their like, chooser, or picker, or whatever.”

Additionally, while metaphors were seen as possibly too complex for children, they were considered possibly beneficial for educating caregivers and professionals as metaphors may provide relatable examples that enhance understanding.

Modeling and Physical Supports

Focus group members discussed modeling and physical supports that could possibly be used to cue users, model device control, and reinforce language connections. They described the possible benefit of physically pointing to the target item to guiding the user’s attention to the item they should be focusing on and help reinforce language concepts. In this regard, P5 said, “a ton of input so that they learn what these symbols mean and the language concepts behind them mean. And again, I would probably do it pointing.” Another method highlighted by participants was tapping the individuals shoulder each time the target item flashed to help build an association and reinforce attention to target item highlighting. A participant also noted this may elicit a somatosensory P300, which may support elicitation of the P300 brain wave. Finally, the practice of turn-taking between the individual using the BCI-AAC and the trainer was discussed, with the BCI-AAC trainer thinking aloud what they were doing to control the device to reinforce BCI-AAC control methods. In completing this turn-taking task, BCI-AAC control could be simulated, meaning the trainer may not have to wear the EEG cap.

Providing Feedback

Focus group participants emphasized the importance of feedback, highlighting the role of immediate and contingent feedback from conversation partners in reinforcing positive behaviors and helping users understand how to control the BCI-AAC system. They discussed letting users know precisely what they did that may have helped support device control. The use of natural consequences was also discussed to help users understand the impacts of making BCI-AAC selections. For instance, P1 said “Yeah, if we have a little go, icon, and that child happens to make it say, go. I’m just saying well, you said go, and we’re moving and, and, reinforcing the action.” Additionally, participants discussed the application of biofeedback, specifically utilizing alpha wave brain activity, to provide users with feedback. By monitoring levels of alpha waves, which indicate varying states of attention, instructors can offer feedback that reflects the user’s attention level during the BCI-AAC task. Finally, having the display item change in color or size as the BCI-AAC algorithm moves toward selection so the user can understand what the BCI-AAC is “thinking” (e.g., the item gets bigger as the BCI-AAC algorithm progresses in confidence toward selecting it), was also described.

Positioning

The focus group described the significance of body positioning and electrode placement in supporting BCI-AAC performance. They noted the importance of a stable base of support; with users needing to have a strong and stable trunk position while seated. Additionally, the group discussed potential solutions to address challenges related to electrode placement. Issues can arise when electrodes are compressed by factors such as seating or support structures. However, they noted possible engineering solutions, such as flat electrodes or specially designed headrests, to possibly mitigate these challenges.

Considerations for Cortical Visual Impairment

The focus group highlighted the importance of accommodating cortical visual impairment (CVI) in supporting BCI-AAC use. Regarding CVI, P6 stated:

“And I think so many of our kids with physical issues, especially cerebral palsy. You always have to have cortical visual impairment, cerebral visual impairment in the back of your head. And so that really impacts lighting and positioning of things. And and so that’s something that I think always has to be front of mind with these guys.”

Discussion

The findings of this foundational study identify several factors that may support BCI-AAC training for a diverse group of children who might use P300-BCI-AAC. The discussion is organized to parallel the results for clarity.

Scaffolding

A significant portion of the focus group emphasized the importance of scaffolding. Scaffolding is a widely used technique in AAC (McNaughton et al., 2008) to develop proficiency in AAC use. Scaffolding supports AAC success by creating flexible and adaptive systems that match each child’s developmental level and trajectory, ensuring that the child does not need to adapt to the BCI-AAC technology. Interface customization is an important concept for both AAC and BCI-AAC implementation to enhance motivation, interest, and support individuals varying skills (Pitt et al., 2022; 2024b). Participants suggested various supports to help scaffold individual competence with P300-BCI-AAC. They highlighted the potential need for pre-teaching related skills for BCI-AAC control, such as incorporating attentional tasks with physical objects or natural activities (e.g., attending to a TV commercial). They also noted the benefit of allowing the BCI-AAC trainer to have manual control over selection accuracy to speed up the selection process, reinforce desired behaviors, and boost the user’s confidence. For instance, trainers could use a mouse cursor to select which item the BCI-AAC chooses and when.

Participants discussed a range of visual display adaptations and simplifications. Similar to Pitt et al. (2019b; 2022), they suggested using game-based training activities like “whack-a-mole” and those games developed by developers such as Look to Learn and Timocco. These games may engage individuals and lower language burdens. For instance, since sustained attention to a target item is required for P300-BCI-AAC use, a game may only respond (the mole is whacked) after a specified duration, which may be increased to teach the needed attentional behavior. They also recommended providing initial exposure to the likely unfamiliar flashing interface without expecting the individual to make an item selection, promoting familiarity. However, participants emphasized the need for communication and social interaction opportunities, choosing fun and motivating tasks that match the child’s interests and clearly connect BCI-AAC use with real-world communication.

Methods for guiding attention to the target item to help users differentiate between engaging with the target item and ignoring distractors were discussed. Options included using something fun, interesting, and preferred (e.g., a picture of a favorite toy or close family member), or items that move, change color, or make engaging or “silly” sounds. These suggestions align with existing BCI-AAC and AAC literature. For example, using color (Ryan et al., 2017), and animation (Pitt et al., 2024a), can increase attention to items within the P300-BCI-AAC display, and preferred items are commonly used to help build intentional communication (Beukelman & Light, 2020). Engaging sounds could also possibly facilitate AAC control and language learning (Boster et al., 2024). Slowing down the flash speed was also considered to bolster attention to the target. A decreased rate, possibly with a pause feature, could provide time for processing and reinforcement of instruction. Additionally, using a physical symbol, possibly during this pause time, could help reinforce the connection with the symbol on the computer display.

Participants also suggested reducing display complexity by decreasing the number of items in the array or concealing them with a physical object, such as a fly swatter, to enhance focus on the target item. However, it is important to consider that the rarity of the target item among non-targets impacts P300 ERP elicitation, with rarer targets generating a bigger response (2019a). Therefore, while small (e.g., 6 item arrays; Pitt et al., 2024a), may elicit a P300-ERP, the effects of reduced display sizes on the P300 signal and possible compensation methods (e.g., manual control over the item selector) should be considered. The environment was also discussed regarding simplification, emphasizing the importance of minimizing environmental distractions during learning.

While participants discussed various scaffolding strategies, such as differing interface adaptations, it should be considered that the optimal sequence for their implementation remains unclear. Further research is needed to determine how the different scaffolding techniques discussed can be effectively combined and tailored to support BCI-AAC use. Furthermore, how scaffolded supports may be faded over time to promote independent BCI-AAC use requires consideration. It is also of note to consider how methods such as independent exploration, imaginary play, drill and practice, and structured instruction could be used during scaffolded training to enhance success (McNaughton et al., 2008).

Verbal Instruction Along with Modeling and Physical Supports

The focus group discussions highlighted several strategies to enhance BCI-AAC training for children, describing verbal instructions and modeling and physical supports, which may be combined to align with current clinical practices to support AAC success (Wandin et al., 2023). Participants emphasized the importance of using explicit, concrete, and simple language, along with allowing additional processing time to enhance the clarity of instructions for BCI-AAC. The integration of BCI with existing AAC methods was also discussed. While the benefits of integrating BCI with existing technologies has been previously acknowledged (e.g., Pitt et al., 2022), participants specifically highlighted the value of leveraging familiar instructions that the individual has previously found useful in learning a prior AAC method to promote familiarity with the new BCI-AAC system. Metaphors were also suggested as a tool to enhance understanding by linking new concepts to familiar schemas. For example, the analogy of a game of tag was used by Light et al., (1993) to support scanning-based AAC learning by helping coordinate attention between the cursor (analogous to “it” in tag) and the target item (the one to be “tagged”). However, participants noted that metaphors might be too complex for children with limited language skills and might be better suited for educating caregivers and professionals about AAC technology. Beyond metaphors, participants supported existing research by suggesting spoken prompts when the target item is highlighted, such as saying “now” (Huggins et al., 2022), or instructing the individual to count the number of flashes, when appropriate to their skills.

Modeling and demonstrating BCI-AAC use were recognized as a priority for SLPs to facilitate intervention (Pitt et al., 2024b). Although P300-BCI-AAC control does not require physical movement, participants considered various ways to demonstrate its use effectively. Similar to current AAC instruction methods for access techniques such as eye gaze (Wandin et al., 2023), participants described pointing to symbols on the computer screen to reinforce language connections and guide the user’s attention. They also discussed tapping the user’s shoulder each time the item flashed to cue the child. The P300 response can be elicited through visual, auditory, and tactile means (Pitt et al., 2019a). Therefore, this tactile cue may evoke a P300 response, which could enhance early BCI-AAC performance, particularly if the child is not generating a strong P300 in response to visual stimuli. However, the potential effects of this strategy should be carefully evaluated as the child progresses, since differences in timing could cause the tactile P300 to interfere with or obscure the visual P300. Participants also emphasized the role of turn-taking in demonstrating BCI-AAC performance. In this approach, the user controls, or pretends to control, the BCI-AAC system, thinking aloud about their actions to reinforce control methods. It is possible a simulation of BCI-AAC control could be done without the trainer wearing the EEG cap, making it a more flexible training method. However, careful consideration is needed to avoid encouraging vocalizations or verbalizations from the child using the BCI-AAC when making selections, as these can create muscular artifacts that degrade the EEG signal and impair performance. Further, it is possible that some children may benefit from their communication partner also wearing an EEG cap, even if it is not connected to the system, as this could help maintain consistency with the child’s own BCI experience, promoting a sense of familiarity during interactions.

Providing Feedback

The role of feedback in supporting language acquisition, AAC, and BCI-AAC competence is important for a multitude of reasons, including reinforcing control strategies and helping individuals understand how they can impact their environment. Focus group participants highlighted the importance of immediate and contingent feedback to users, reinforcing positive behaviors and helping them understand the consequences of their actions. In addition to aiding in the establishment of initial P300-BCI-AAC use, immediate reinforcement of user behaviors may also be important for maintaining the individual’s sense of agency as their performance with the system improves. The use of natural consequences, a commonly utilized AAC approach, was also underscored to help users understand BCI-AAC control and its impact on their surroundings, reinforcing BCI control and communication.

Participants emphasized the critical importance of the communication partner in delivering immediate and responsive feedback to P300-BCI-AAC users. Additionally, EEG methods were identified as a potential feedback mechanism. For instance, prior work has provided initial support for using a brain wave known as alpha to provide feedback. Alpha wave brain activity represents factors such as an individual’s engagement level with a task. Therefore, building on the work of Galvin-McLaughlin et al., (2022), alpha levels may potentially be used to provide a child with some form of feedback (e.g., a happy face), when levels of attention are higher, and another form of feedback (e.g., a sad face), when attention is lowered. Further, feedback may be given to the user by changing the item’s appearance (e.g., movement, color) as the BCI algorithm progresses toward making a selection. For example, similar to existing eye-gaze methods where a circle slowly forms as the user gazes at an item to make a selection, Pitt & Brumberg (2022) incorporated a slowly forming circle in their scanning-based BCI interface to help the user understand what item the BCI “thinks” they are trying to select. Further research is needed to evaluate these EEG-based feedback approaches in children.

Positioning and Considering those with CVI

The focus group underscored the importance of positioning for AAC users. Without optimal positioning, users may experience increased fatigue and tiredness, which can negatively impact their attentional function and their ability to optimally visualize the communication display (Beukelman & Light, 2020; Sowers, & Wilkinson, 2023). Proper positioning is also important to limit primitive reflexes, such as the asymmetrical tonic neck reflex, which could interfere with BCI-AAC control by causing muscle activations that degrade EEG signal quality (Brumberg et al., 2018). Therefore, it is unsurprising that participants emphasized the importance of ensuring a strong and stable trunk support and being at an appropriate angle to support attention towards items in the P300-BCI-AAC system. They also noted that compression of EEG recording electrodes can decrease signal quality and, consequently, BCI-AAC outcomes. Therefore, engineering solutions such as flat electrodes or customized headrests may be necessary to optimize EEG recordings while maintaining optimal user positioning.

Finally, it was highlighted in the focus group that there is a need for awareness and consideration of individuals with CVI, a condition that affects the brain’s ability to interpret visual information and is common among those with physical disabilities (Wilkinson et al., 2023). Optimal methods for supporting access and training for individuals with CVI may include the use of auditory stimuli (e.g., Boster et al., 2024), tactile stimuli, larger symbols, high contrast, and animation (Boster et al., in review). While these suggestions align with current AAC and BCI-AAC practices, further research is necessary to support the growing body of research in this area (e.g., Blackstone et al., 2022; Boster et al., in review; 2021; 2024; Wilkinson et al., 2023) and expand on these methods to effectively accommodate users with CVI.

Limitations and Future Directions

The study has several limitations that warrant consideration and suggest directions for future research. First, while the focus groups provided valuable insights into BCI-AAC training strategies, it included a limited sample of individuals some of which has limited BCI-AAC experience. Further, the study primarily focused on the perspectives of experts in AAC and BCI-AAC with a background in SLP. SLPs were primarily included because of their integral role in prescribing AAC systems, supporting their use, and to help promote the systematic evaluation of stakeholder perspectives. However, further research should evaluate the perspectives of a greater range of professionals from diverse fields (e.g., occupational therapy, physical therapy). Further, future research should seek to include AAC users and their families. Systematically including a broader range of stakeholders will help provide a more comprehensive understanding of training needs and practical challenges. Additionally, the study’s focus was primarily on P300-based BCI-AAC systems. Future research should explore training strategies for other types of BCI, such as those based on motor imagery or other neural signal modalities This broader approach could uncover unique challenges and effective training methods pertinent to different BCI technologies. For further reading on different BCI-AAC techniques beyond the scope of this work, the reader is referred to Brumberg et al., (2018) and Pitt et al., (2019a). Finally, the study did not address the long-term effectiveness of the recommended training strategies or how these methods might be adapted for different developmental stages or changing needs over time. Future research should consider longitudinal studies to evaluate the sustained impact of various training methods and adapt strategies as users progress or as new BCI technologies emerge. This comprehensive approach could enhance the practical applicability and effectiveness of BCI-AAC training across diverse user populations.

Conclusion

This study provides valuable, early, insights into effective strategies for training children in P300-BCI-AAC systems through qualitative focus groups. Utilizing a diverse array of participant experiences, directions for P300-BCI-AAC training are considered that align with current AAC practices to promote possible integration. Participants discussed a range of topics including scaffolding techniques, verbal instruction, modeling and physical supports, feedback mechanisms, positioning and application to those with CVI. Our findings emphasize the importance of tailoring training methods to the unique developmental levels and needs of each child. Future research should expand to include a broader range of stakeholders, explore different BCI technologies, and evaluate the long-term effectiveness of training strategies. This comprehensive approach may help support BCI-AAC training and accessibility for a diverse population of children with communication and physical challenges. However, as a foundational study, the approaches discussed likely only begin to address the complexities of using BCI-AAC with children. Future research is needed to expand on this work.