Ethical Boundaries in Ultra-Fast Neurotechnology
Where Does Human Agency End?
Establishing timing thresholds to preserve human sense of agency, autonomy, and legal responsibility in brain-computer interface systems that can act before you do.
Introduction
Voluntary Action and Human Agency: Bridging Philosophy, Neuroscience, and Human-Brain Interface
Throughout history, the nature of human action and our abilities as agents have been central themes in philosophy. The concept of agency is intrinsically linked to the idea of human autonomy and the extent to which individuals can exert control over their actions. This fundamental inquiry, dating back to the early days of philosophical thought, has driven investigations into human consciousness and self-awareness.
Leibniz's "Theodicy" exemplifies this line of inquiry, posing the problem of necessity and questioning whether humans are truly free agents or merely subject to the determinism of a pre-established order. This understanding of agency presupposes individuals' ability to identify themselves as agents capable of initiating and controlling actions. Such self-awareness is fundamental to human nature, permitting purposeful behavior and decision-making that reflects individuals' intentions, desires, and beliefs.
Transitioning from traditional philosophical debates into the realm of contemporary neuroscience research, voluntary action is depicted as entailing two distinct subjective experiences: intention and the sense of agency (SoA). The concept of intention primarily refers to the cognitive processes underlying the planning and initiation of actions. It enters into the vast and often contested discussion about human freedom in decision-making and is closely tied to the concept of 'free will', a term we consciously avoid here to prevent getting entangled in complex philosophical debates. While intention is typically probed via introspective methods, it has been empirically investigated through various experimental designs as well.
Contrarily, SoA deals with the subjective experience of controlling one's actions and perceiving oneself as the author of those actions. Unlike intention, the SoA is often studied retrospectively, examining how individuals perceive their actions after they have occurred. This distinction helps us better channel our research efforts to explore the unique aspects of each concept and ensure that our investigations remain grounded within their respective theoretical frameworks.
Focusing on SoA, our research aims to delve deeper into the subjective experience associated with the execution of a voluntary action. The exploration of SoA holds promise in illuminating the human condition, our freedom, our responsibilities, and our control over actions. Probing the neural and cognitive mechanisms underpinning SoA, we aspire to demystify the processes enabling us to experience ourselves as agents.
The Role of SoA in Ethics and Law
The SoA also takes center stage in the realms of ethics and law, particularly in the context of emerging technologies like brain-computer interfaces (BCIs). Deciphering responsibility for actions performed through these devices raises urgent questions. These investigations collectively highlight SoA's potential to bolster human performance while grappling with the ethical complexities spurred by technological advancement.
Theoretical Foundation: This research builds upon concepts from: Ghane, H. (2023). The Point of No Return in Action Cancellation: Deciphering Its Influence on the Human Sense of Agency via Real-Time Brain-Computer Interfaces. Master's Thesis, Pompeu Fabra University. View Thesis
Narrowing the Concept of SoA
Our investigation of SoA is more specifically oriented. Whereas some definitions interpret SoA as a broad subjective feeling of capability to act (or self-efficacy), we take a more restricted approach. We center our exploration on the experiential component of SoA, which emerges before, during, and after an actual muscular movement. Therefore, throughout this work, SoA refers to the subjective experience related to the execution of a particular motor action.
Actions form the core of our everyday experiences and are fundamental for achieving our goals. Human actions can generally be divided into two categories: stimulus-dependent reactions and stimulus-independent voluntary actions. Reactions are direct responses to external stimuli, while voluntary actions are internally initiated, uninfluenced by external stimuli, and accompanied by unique subjective experiences like the sensation of personal choice.
For this work, we adopt a pragmatic criterion: we deem an action voluntary if no apparent, discernible external stimulus perceived by an average human triggers the action. This approach provides a working definition of voluntary actions, enabling us to investigate SoA in a more structured and focused manner.
The Problem
Why Now?
We've entered an era of neurotechnology and ultra-fast HMIs where machines can act almost as soon as our minds lean toward action. Like it or not, this nudges us toward an extended self.
To maintain autonomy and trust, we need explicit, straightforward protocols that clearly delineate the boundary between human responsibility and machine responsibility, both legally and ethically.
SoA-Building Window
When we act, the feeling "I did that" doesn't pop up all at once. It builds:
intend → prepare → (SoA builds — can still cancel) → commit → act
That brief SoA-building window—even at the very last moment—lets the brain check: "Am I still choosing this?" Keeping that option alive is what makes the action feel truly ours.
The Neurotech Problem
Some of today's fast HMIs can start outcomes too early, before this window does its job. Then people report weaker agency; autonomy, consent, and responsibility get shaky—and disputes become harder to resolve.
We need a clear, testable timing rule so builders and lawyers can draw the human/machine line.
Research Approach
Hypothesis & Aim
Hypothesis: If an outcome is started too early after intention detection, later sense of agency drops; there exists an earliest safe time that preserves SoA.
Aim: Estimate this SoA-building window and turn it into a simple design rule and timing log.
Experiment 1: Finding the Critical Threshold
Design
- Detect early intention in real-time (e.g., keypress task)
- Start a brief sound before the action at different times across trials
- After each trial, ask "How much did it feel like you did that?"
- Log three times: intention detected, outcome started, action performed
Expected Outcome
From timing vs. ratings, find the earliest safe time → this estimates the SoA-building window.
This threshold defines where human agency is preserved vs. compromised.
Experimental setup showing intention detection, feature extraction, and varying pre-action outcome timing
Experiment 2: Temporal Recalibration
Hypothesis
When participants exhibit a SoA for outcomes preceding their actions, it suggests that temporal recalibration has realigned the perceived timing of the action and outcome.
This realignment ensures the preservation of the causal relationship between one's action and subsequent outcome.
Method
- Detect intention based on specific HMI protocol
- Initiate outcome before action (after threshold from Exp. 1)
- Measure perceived time of outcome and action
- Investigate if brains recalibrate timing to preserve causality
Diagram showing how perceived timing of action and outcome may cross, establishing causal temporal recalibration
Voluntary action framework showing prospective (pre-action) and retrospective (post-action) agency
Research Implications
1. Temporal Convergence of SoA and JoA
Traditionally, the sense of agency (SoA) and judgment of agency (JoA) occur at distinct times. SoA is an immediate feeling tied to actions, while JoA is a reflective assessment of control over outcomes.
Our experimental design, where the outcome is initiated before the action, offers a unique setting to explore their potential overlap—challenging traditional views of these processes being separate in time.
2. Operational Thresholds
Our experiments aim to identify a specific time threshold. Initiating outcomes or actions before this threshold might compromise the user's SoA over the outcome or action.
This has potential legal and ethical ramifications and can be detrimental to user experience. A diminished SoA can lead to user frustration and feelings of detachment or loss of control.
What Changes for Law & Design
Don't start action earlier than the SoA-building window (or always keep hold-to-commit/confirm mechanisms).
If an action starts earlier than the window, responsibility leans toward the system, not the user.
Tell users the device waits long enough for agency to form; flag modes that shorten the wait.
Outputs & Collaboration
Deliverables
- A practical time range (per task/device)
- One-page builder checklist
- Timing log template
- Short policy note for regulators
- Design guidelines for agency-preserving HMIs
Collaboration Opportunities
Open to legal/BCI collaborators to refine standards and deployment.
This research bridges neuroscience, control engineering, ethics, and law to ensure neurotechnology enhances rather than diminishes human autonomy.
"Preserving the user's SoA should be paramount, even if it means imposing limits on the operational speed of these technologies."