From Spikes to Self-Regulation: The Foundations of Neurofeedback

Long before the word neurofeedback became part of our clinical lexicon, the pioneering work of Eberhard Fetz (1969) and Allen Wyler (1976) offered a glimpse into something astonishing: the brain’s ability to learn to control itself, one neuron at a time. In their studies, monkeys were trained—through operant conditioning—to alter the firing rates and patterns of single cortical neurons. Each spike, once thought to be an unchangeable output of internal circuitry, became part of a behavioral repertoire shaped by reinforcement.

These early experiments were invasive, requiring microelectrodes implanted in the precentral motor cortex. Yet they laid the groundwork for what would become the conceptual and experimental foundation of neurofeedback—the process by which individuals learn to modulate their own brain activity through real-time feedback. What began with wires and oscilloscopes in the 1960s would evolve into computer-based EEG and fMRI systems capable of reading and training entire networks.

While these studies could not ethically be replicated in humans today, their spirit lives on in every neurofeedback session conducted in clinics and research labs worldwide. They demonstrated that self-regulation is not an abstract ideal—it’s a fundamental property of neural systems capable of learning.

Methods

In Fetz’s 1969 Science paper, single neurons were recorded from the precentral cortex of awake macaque monkeys. Using an implanted microelectrode system, the animals received food rewards whenever specific neurons increased their firing rate above a threshold. A simple resistor-capacitor integrator converted neural spikes into a running voltage proportional to firing rate, triggering reinforcement when the rate stayed high. Auditory or visual feedback—clicks or meter deflections—helped the animal discriminate when the desired firing rate had been reached.

The monkeys quickly learned to increase activity in individual cells by 50–500% compared to baseline, often generating new motor patterns or subtle wrist movements. Importantly, firing rates dropped back to baseline when reinforcement ceased—proof that learning, not mere arousal or mechanical stimulation, was driving the change.

Building on this, Wyler and Prim (1976) extended operant conditioning to control not only firing rate but firing pattern. Using a paradigm called Differential Reinforcement of Window activity (DRW), they trained monkeys to maintain tonic neuronal activity within predefined interspike intervals (ISIs). The computer-based feedback system reinforced consistent “windowed” firing patterns while penalizing deviations. Statistical analyses confirmed that variance in firing rate decreased significantly across conditioning periods, and in many neurons, the learned pattern persisted even after feedback was withdrawn.

These elegant, if invasive, paradigms proved that the brain could be taught to modulate single-neuron firing—voluntarily, consistently, and with measurable learning curves.

Results

Across both studies, results were nothing short of revolutionary. Fetz’s monkeys learned to voluntarily increase firing of individual cortical neurons—a finding that shattered the long-held belief that single-neuron activity was purely reflexive or automatic. The neurons responded predictably to behavioral contingencies, demonstrating that the central nervous system could integrate reinforcement information to guide activity at the cellular level.

Wyler’s findings extended this control into the temporal domain: monkeys not only increased firing rates but could reshape firing patterns—transitioning from phasic to tonic discharges as reinforcement demanded. Variance and error magnitude in interspike intervals decreased significantly over time, confirming systematic operant control. In several neurons, these learned patterns persisted post-conditioning, suggesting a short-term memory trace of the new activity.

Together, these studies provided empirical evidence that neural activity could be operantly conditioned, setting the stage for contemporary understandings of self-regulation, neuroplasticity, and the neural mechanisms underlying neurofeedback.

Discussion

The combined findings of Fetz (1969) and Wyler & Prim (1976) firmly established that neuronal firing can be influenced through direct reinforcement. In both experiments, changes in activity were not caused by arousal, stimulus-driven responses, or nonspecific excitation, but rather by consistent behavioral contingencies linking firing rate or pattern to reinforcement.

Fetz demonstrated that when food pellets were delivered contingent upon neuronal discharge rate, firing increased significantly above baseline, and decreased again when reinforcement stopped. Control conditions ruled out effects of random reward or sensory feedback, confirming true operant learning at the single-neuron level.

Wyler and Prim built on this by showing that reinforcement could shape patterns of neuronal discharge—specifically, by training monkeys to sustain tonic firing within a narrow interspike interval window. The reduction in firing-rate variance and persistence of new tonic patterns during post-conditioning periods showed measurable, quantifiable control of neural activity. Their use of statistical analysis, including ANOVA and t-tests, strengthened the evidence that learning occurred as a systematic, reproducible process.

Together, these studies demonstrated that neurons in the motor cortex could become the target of operant conditioning, that such control could be objectively measured, and that changes persisted beyond immediate reinforcement. These results remain among the earliest demonstrations of direct behavioral control over neural firing, and they provided the methodological foundation for later work on brain self-regulation and neurofeedback.

Brendan’s Perspective

It’s hard not to feel awe when revisiting Fetz (1969) and Wyler (1976). These were scientists working at the frontier—pushing electrodes into cortex, listening to single spikes, and discovering that neurons could learn. Of course, such invasive experimentation could never be replicated in humans today for ethical reasons. But we now have the tools to revisit their questions safely, and perhaps even more powerfully.

Imagine conducting EEG neurofeedback within an fMRI scanner, tracking not just surface rhythms but deep hemodynamic correlates of learning in real time. We could watch, moment by moment, how cortical regulation spreads across networks—how a trained alpha increase in EEG aligns with activity changes in anterior cingulate, thalamus, or insula. These are the kinds of experiments that could finally satisfy the skeptics who keep asking why neurofeedback works.

Most clinicians, understandably, are focused on outcomes—on whether their clients sleep better, focus longer, or feel calmer. And the good news is: they do. But the mechanistic purists, the “hard-nosed fundamentalists” of neuroscience, want to see the causal chain between reinforcement and reorganization. Thanks to the groundwork laid by these early pioneers, we know that chain exists. What we need now is funding, time, and a bit of patience to trace it in humans, non-invasively, one network at a time.

Until then, the practical work continues. Every time a clinician reinforces a patient’s SMR or alpha band, they’re echoing the same learning principle Fetz demonstrated in 1969: the brain learns to listen to itself—and to change.

Conclusion

The work of Fetz and Wyler transformed neuroscience by proving that the brain’s activity could be shaped through learning. Their invasive studies set the stage for the modern era of non-invasive neurofeedback, translating single-neuron operant conditioning into a system for global self-regulation. With today’s multimodal neuroimaging, we stand at the threshold of understanding not just that neurofeedback works, but how and why it works.

The message from the archives is clear: the brain is not a passive organ—it’s a learner. And we’re only beginning to grasp the depth of what it can teach itself.

References

• Fetz, E. E. (1969). Operant conditioning of cortical unit activity. Science, 163(3870), 955–958.

• Wyler, A. R., & Prim, M. M. (1976). Operant conditioning of tonic neuronal firing rates from single units in monkey motor cortex. Brain Research, 117, 498–502.