The Learning Process in Biofeedback: Feed-Forward or Feedback?

When we think of biofeedback, we often picture an individual sitting in front of a screen, monitoring subtle physiological cues and gradually learning to regulate them. This study, published in 1986 by Dunn, Gillig, Ponsor, and Weil, offers a fascinating window into the learning process behind biofeedback. As an archival study, it reflects the early attempts to disentangle whether biofeedback works through feedback—using signals to construct new responses based on subtle physiological sensations—or feed-forward—drawing on existing strategies and using feedback mainly as confirmation.

Biofeedback, in general, provides real-time information about processes like heart rate, muscle activity, or skin conductance, which are usually outside conscious awareness. Neurofeedback, a subset of biofeedback, focuses specifically on brain activity measured via EEG. Both methods have been applied to support self-regulation, from reducing muscle tension and chronic pain to improving attention and emotional regulation. But how exactly do people learn to use the information? The answer has implications not only for refining protocols but also for ensuring that the skills transfer into daily life.

This archival article is especially interesting today, as biofeedback and neurofeedback continue to evolve with new technologies. Revisiting these foundational questions reminds us that the heart of training lies not just in the data displayed but in how individuals engage cognitively and behaviorally with the process.

Methods

Twenty upper-level undergraduate students (15 women and 5 men) participated in three electromyograph (EMG) biofeedback sessions. The sessions targeted muscle tension in the forehead, measured using an Autogen 1700 EMG device. Active electrodes were placed on the forehead, and participants received auditory feedback in the form of a tone. The task was simple: try to keep the sound off as much as possible. The tone’s volume decreased with muscle relaxation, and cut off entirely when participants reduced tension below a set threshold.

Structure of Sessions

• Session I:

  • 5-minute baseline.

  • Two 10-minute biofeedback runs at a threshold set 25% below baseline.

  • Participants were asked about their strategies and thoughts during the session.

• Session II:

  • 5-minute baseline.

  • One 20-minute biofeedback run at 25% below baseline.

  • Verbal reports of strategies after training.

• Session III:

  • 5-minute baseline.

  • A 5-minute reversal run where feedback was inverted: the sound switched off when tension increased (unknown to participants).

  • A 15-minute return-to-normal biofeedback run at 25% below baseline.

Data Collected

1. Physiological: EMG levels (mean of last 5 minutes per session).

2. Cognitive: Anticipatory, concurrent, and retrospective verbal reports were recorded and analyzed.

Analysis

Judges evaluated the verbal protocols against four criteria:

• A: Constructing a new response (feedback).

• B: Trying a known response (feed-forward).

• C: Using feedback to sense subtle muscle differences (feedback).

• D: Using feedback as confirmation of a response (feed-forward).

Ratings (agree, disagree, unsure) allowed researchers to determine whether feedback or feed-forward processes predominated.

Results

• EMG Reductions: Across all sessions, participants showed lower muscle tension during biofeedback compared to baseline. However, performance did not improve systematically over sessions.

• Verbal Protocols: Judges consistently found more evidence for feed-forward strategies (items B and D). Participants were more often described as drawing on existing strategies and using feedback only as confirmation, rather than constructing new strategies from subtle sensations.

• Reversal Condition: In Session III, 11 of 20 participants noticed something unusual during the reversed feedback, but only 2 correctly identified the change in contingency.

• Overall: The majority of learning was interpreted as feed-forward, with little evidence that participants developed fine-grained sensitivity to subtle changes in muscle tension.

Discussion

This archival study raises enduring questions about how people learn in biofeedback. The evidence suggests that participants relied primarily on feed-forward processes: they tried strategies they already knew (e.g., “relaxing,” “thinking calming thoughts”) and checked the feedback signal for confirmation. Few participants reported constructing entirely new responses based on nuanced awareness of muscle sensations.

Why does this matter? If biofeedback learning relies heavily on feed-forward processes, there may be limitations:

• Reduced fine control: Without tuning into subtle sensations, individuals may not fully optimize their responses.

• Weaker transfer: Skills may not fully generalize beyond the training context, since participants did not learn to detect and regulate tension without external cues.

This resonates with ongoing challenges in both biofeedback and neurofeedback practice today: How do we ensure that skills extend into real life? What role should protocols and coaching play in fostering deeper awareness rather than superficial strategy use?

Importantly, while this study emphasized feed-forward dominance, the authors acknowledged that with longer or more varied training, feedback-driven processes might still emerge. However, subsequent research has often confirmed that participants rely more on explicit strategies than on subtle physiological awareness—unless carefully guided otherwise.

Brendan’s Perspective

Looking at this study from today’s vantage point, it feels like a time capsule of the challenges we still face in clinical practice. Clients often arrive at neurofeedback or biofeedback sessions with a toolbox of coping strategies—breathing, visualization, distraction—and they naturally try these first. The data from this 1986 study remind us that unless we design protocols and provide guidance that draw clients’ attention inward, they may never truly learn to sense and refine those subtle bodily cues that make biofeedback transformative.

In practice, this means:

• Sometimes using continuous feedback rather than simple threshold cutoffs, so clients receive more nuanced information. As this theoretially leads to weaker conditioning, it’s important to use this during the various “psychoeducation” portions of a training protocol. 

• Incorporating guided awareness exercises, helping clients connect physiological signals with internal sensations.

• Applying protocol individualization, especially in EEG neurofeedback, where training frequencies and electrode placements (e.g., SMR training at C3/C4 for relaxation, alpha enhancement at Pz for stress reduction) can be tailored to maximize awareness and self-regulation.

• Encouraging integration into daily life, where clients practice sensing shifts in muscle tension, breath, or brainwave states outside of sessions, gradually reducing reliance on the equipment. Biofeedback is a great bridge for neurofeedback transfer, and it is one of my “cheat codes” for neurofeedback that you absolutely must use. 

This article also illustrates the pitfalls of early research designs: short training (only three sessions), reliance on student participants, and limited outcome measures. In real clinical contexts, training often spans 20–40 sessions, allowing the gradual layering of awareness, control, and transfer. The difference between experimental biofeedback and clinical biofeedback is, in many ways, the difference between feed-forward reliance and genuine feedback learning. Would the participants in this study have naturally shifted strategies over time? Or is a qualified neurofeedback practitioner a necessary part of that aspect of neurofeedback training? 

For modern neurofeedback professionals, the lesson is clear: design for awareness, not just performance. When protocols emphasize subtle discrimination—whether in muscle relaxation, heart rate variability, or EEG rhythms—clients learn skills that stick. The enduring message from this 1986 study is that feedback must be more than confirmation. It must become a teacher.

Conclusion

This archival research sheds light on a fundamental tension in biofeedback: are we truly teaching new skills, or simply confirming existing strategies? The findings leaned heavily toward the latter, with participants relying on feed-forward processes. Yet the clinical implications are profound. To make biofeedback and neurofeedback effective, we must cultivate conditions that promote subtle awareness, deep learning, and real-world transfer.

The ultimate take-home message: biofeedback is most powerful when it shifts from confirmation to transformation—helping individuals discover and refine new ways of sensing and regulating their bodies and minds.

References

Dunn, T. G., Gillig, S. E., Ponsor, S. E., Weil, N., & Utz, S. W. (1986). The learning process in biofeedback: Is it feed-forward or feedback? Biofeedback and Self-Regulation, 11(2), 143–156. https://doi.org/10.1007/BF00999738