The Numb UI: Why 100ms is the Speed Limit of Human Agency
Exploring the neurobiology of computer interaction—why latency is a fundamental design constraint for maintaining a user's sense of control.
In my first post on this blog, I talked about a feeling of disconnection and numbness. At the time, I was speaking emotionally, but as I've spent more time deep in the technical weeds of Edge Computing and Local-First Architecture, I've come to realize that "numbness" isn't just a mood. It's an architectural byproduct of modern web development.
When we build applications that rely on distant servers for every interaction, we aren't just introducing delay. We are breaking the fundamental neurobiological feedback loop that allows humans to feel in control of their tools.
We are building "Numb UIs."
1. The Perceived Causality Threshold
In 1968, research scientist Robert B. Miller published a seminal paper on response times in human-computer transactions. He identified a critical threshold: 100 milliseconds.
This isn't an arbitrary number. It is the approximate limit of the human brain's perception of "instantaneous" feedback. When an action (moving a mouse, typing a key) is followed by a reaction within 100ms, the brain perceives a direct causal link. We feel like we are moving the pixels on the screen.
Beyond 100ms, the illusion of causality begins to fracture. The computer stops feeling like an extension of our body and starts feeling like an external agent that we are "requesting" things from. We move from a state of "flow" to a state of "transaction."
2. The Science of the "Sense of Agency"
This phenomenon has a name in cognitive neuroscience: the Sense of Agency (SoA). It is the subjective experience of being the author of one's own actions and their consequences.
Seminal studies by Haggard et al. (2002) have shown that when we perceive ourselves as the cause of an event, our brain subjectively "compresses" the time between the action and the result. This is known as Intentional Binding.
However, intentional binding only works within a tight temporal window. If the feedback from an application is delayed by 200ms or 500ms—even if we use "Optimistic UI" updates—the brain's action-control network (including the bilateral anterior insular cortex) registers a mismatch. This mismatch triggers an "orienting response," pulling us out of our task and forcing us to consciously acknowledge the system's latency.
This is exactly what contributes to the Cognitive Load I discussed recently. Every millisecond of delay is a mental tax that prevents us from entering a true state of "Deep Work."
3. Why Skeleton Screens are a Lie
In recent years, the industry has leaned heavily on "perceived performance" techniques. We use skeleton screens, loading spinners, and progressive blurring to hide latency.
But these are purely visual bandaids for a neurological problem. Your conscious brain (System 2) might see a skeleton loader and think, "The data is coming," but your subconscious brain (System 1) has already registered the break in causality. The interface still feels "numb" because it hasn't responded to your intent instantly.
This is why Local-First Software is the only real solution. By running a database like SQLite directly in the browser, we can guarantee sub-50ms response times for nearly every interaction. We aren't just making the app faster; we are restoring the user's sense of agency.
4. Building for the Edge of the Self
In 2026, the competitive advantage isn't just about features. It's about Resonance.
Applications that respect the 100ms threshold feel "alive." They feel like an extension of the user's thought process. This is the goal of my LMS project—to create a learning environment where the technology disappears entirely, leaving only the student and the content.
To achieve this, we must:
- Minimize the Network Hook: Stop using the server as a state management tool.
- Leverage the User's Edge: Use the compute power on the device to handle interactions.
- Respect Biological Limits: Understand that 100ms is not a goal; it's a hard limit.
When we build software that respects our biology, we stop building "numb" tools and start building instruments that amplify human potential.
This article is part of an ongoing series on Intentional Engineering. To understand the foundations of this approach, read about Software Minimalism.
