Human Perception Is Also Prompt Based

There Is Something There

We’ve all had the experience: you just watched a horror film.

Now everything you see doesn’t seem very natural anymore. At the corner of your eyes, you start seeing something that’s not there. As if there’s something out there, only starting to show up right after you’ve watched the horror film. We naturally see patterns in chaos—that’s not news—but we can all agree the peak of these hallucinations happens right after we are exposed to a certain scenario.

Even less noticeable are scenarios of the natural kind, in terms of stories, such as romcoms, adventure, or action films. They prime us to perceive the world with a bias toward the information presented in the story.

This led me to believe that our own perception is also prompt-based.

Diffusion Models

The diffusion model (first described in detail in “Denoising Diffusion Probabilistic Models”), if you don’t know what it means, is essentially a process for AI to come up with useful results out of nothing but random noise.

The process roughly goes like this, in the case of image generation:

1. Start with the machine generating a full canvas of noise.
2. You tell the AI what you want the final result to be in words: the prompt.
3. It starts experimenting with activating neurones in the AI model, which in turn creates slightly different noisy patterns that, in one way or another, get closer and closer to what you described in the prompt.
4. After the AI reaches a point of maximum iteration number, or when it thinks the difference between the result and your prompt is small enough, it stops and gives you the final result.

Human Perception As A Diffusion Model

Our perception is awfully flawed. Our vision system, for instance, which we might think is super perceptive, is still very limited. Only the core of our vision is able to discern information in high definition, while the rest of the field of vision is just blurry and sparse pixels.

We don’t feel that way, do we? If you force yourself to “see” things with peripheral vision, you start to realise you can’t actually see much.

But why do we feel like we have a full vision, including the parts where we can’t actually perceive? The reason is our brain is constantly filling in the lost information. Even when we’re most sober, there is still bias, distortion, or inclination based on who we are and where we are at, in any given moment.

This brings up my point: our perception system is actually prompt-based as well, similar to the diffusion model. We are seeing what we want to see or what our brain thinks we are most likely seeing.

This is why, given the same amount of information (or lack of it), we feel differently when our mentality shifts. When we feel happy, we perceive the surroundings to be more positive; when we feel anxious, triggers and alarms spring up from all over the place. Psychological factors aside, I suspect there might be subtle alterations in reality, or at least our interpretation of reality, based on our internal prompts.

Back to the example at the beginning: right after we are exposed to terrifying stories, we tend to make up non-existent objects from our peripheral vision because the prompt is in our short-term memory and is dominant in the active interpretation process running in the background all the time.

Real World Examples

Although probably no one has put it like this, the principle has been used in practice in many areas of our lives.

Advertising

When you think about influencing people toward a specific goal, nothing is better than studying advertising. From something as subtle as product placement in films and TV to the flashy banner ads you see on websites, all ads, at their core, aim to prime and lead you into the funnel: awareness, consideration, and eventually conversion into a paying customer.

Priming is such an important tool that a large portion of advertising budgets actually goes to this stage, especially for well-known brands in the consumer or commodity sectors.

Take, for instance, the Red Bull racing experiment conducted by researchers at Boston College’s Carroll School of Management: participants were asked to play a racing video game where the only difference between cars was the brand logo displayed. Those who drove cars with the Red Bull logo exhibited notably more aggressive and risky behaviour compared to those driving cars with other brands like Coca-Cola, Guinness, or Tropicana. The intriguing aspect? The cars had identical performance capabilities—the only variable was the logo.

This experiment illustrates how brand priming can subtly influence behaviour at a subconscious level. The participants driving the Red Bull-branded cars were unaware of their altered racing style—some excelled with their aggressive approach, while others crashed due to excessive risk-taking. This behavioural shift was rooted in the Red Bull brand’s carefully crafted identity of extreme sports and edgy marketing, highlighting how deeply brand prompts can shape our actions without our conscious awareness.

In both cases, these stimuli guide you through chaos and noise, helping you focus not on what’s random, but what feels aligned with your primed perception. Your hand reaches for the product, seemingly on its own.

Self-Affirmation

Saying nice things to yourself in front of the mirror every day is a proven technique backed by self-affirmation theory to solve many psychological issues.

• You are the best version of yourself.
• You are beautiful.
• You can achieve X, Y, and Z.

These mantras are essentially short prompts for our minds, preparing us to tune in to the correct frequencies amid everyday challenges. The power of such prompting on behaviour has been demonstrated in various studies, including the famous “Florida effect” experiment. In this study, participants who were exposed to words associated with elderly people (like “Florida,” “old,” and “retired”) unconsciously began walking more slowly afterward—showing how even subtle prompts can shape our actions without our awareness.

When faced with a difficulty, a poorly primed mind might perceive it as overwhelmingly destructive, halting progress and deflating confidence. A well-primed person, however, might interpret the difficulty as something positive—perhaps a chance to test out a theory, an opportunity to find flaws, or a passage of adventure that shakes out competition.

Whether through short-term prompts (motivational quotes, group support) or long-term prompting (education, self-affirmation), properly primed people are more resilient during rough patches and better able to find meaning in the presence of hardship.

We shape ourselves in response to life’s ups and downs, and the prompts affect how we perceive and make decisions. In a way, prompt-engineering ourselves is probably worth exploring further. From interacting with AI systems, we know that the quality of input prompts determines the quality of the output.

Memory As Prompt

Memories are not real—well, they are real in a way, but not in the present. As shown in Elizabeth Loftus’s fascinating work on false memories, what has happened is in the past, and it exists only in its consequences and in our minds. It shapes our current decisions, sometimes for better, sometimes for worse. The past is not reality, and in most cases, our memories don’t even fully represent what actually happened. Our emotions and biases tend to modify events to form a cohesive narrative. And that’s exactly what memories are: narratives that help us organise and make sense of the chaotic world.

Memory itself is a form of prompting. AI models use this concept too. When given a series of memories, the model shapes its responses to align with those memories, creating relevance and continuity.

Of course, memories exist on both a personal and collective level. They provide a large contextual prompt for what we do, allowing us to avoid starting from scratch and instead build on what we already “know.”

Emotion As Prompt

How about emotions? I think they are also acting as prompts, perhaps some of the most powerful ones we experience. Our emotional state serves as a filter through which we perceive and interpret the world around us.

When we’re anxious, this emotional prompt leads us to:

• Interpret neutral facial expressions as threatening
• Notice more potential dangers in our environment
• Focus on negative possibilities in future scenarios

Conversely, when we’re in a positive emotional state:

• We’re more likely to see opportunities rather than obstacles
• Social interactions are interpreted more favorably
• Our threshold for stress and frustration increases

Research in affective neuroscience supports this view. Studies have shown that the amygdala—our emotional processing center—can influence visual processing as early as 80 milliseconds after stimulus onset. This suggests that our emotions aren’t just reactions to our perceptions, but actually shape how we perceive things in the first place.

Consider the phenomenon of “emotional contagion”, where we unconsciously mirror the emotions of those around us. This could be viewed as a form of social prompt that influences not just our own emotional state, but our entire perception of the social environment.

Even physical sensations can be reinterpreted based on our emotional prompts. The same physiological arousal (increased heart rate, sweating) might be interpreted as excitement when we’re happy, but as anxiety when we’re stressed—demonstrating how emotional prompts guide our interpretation of bodily signals.

Why Do I Write This?

It happened exactly like the opening: I watched a horror film, and then I started seeing things. The fact that this happens right after watching the horror film is what makes it so fascinating to me. It’s as if I’ve unlocked a dormant capacity—one I haven’t used in a long time. Because of the recent exposure to unsettling images, my short-term mental prompt has shifted. Now, I’m interpreting the randomness and noise around me based on this newly set prompt.

Most people are familiar with this effect—feeling spooked and hyper-aware after consuming horror media. But I think there’s something deeper hiding beneath this phenomenon. This must be how our brains fundamentally work, or potentially how any neurological systems operate (even artificial ones), regardless of their strength or complexity.

Our perception systems are inherently designed to respond to prompts, even in subtle or seemingly insignificant scenarios. This raises an intriguing possibility: could we consciously harness this mechanism as a tool for achieving our goals?

For instance, if the brain is so ready to process prompts, perhaps we can design ways to prime ourselves to interpret randomness and chaos in more constructive, positive ways. Just as exposure to horror primes us to see shadows as threats, could deliberate exposure to certain stimuli prime us for confidence, creativity, or problem-solving?

The AI Connection

Another question lingers in my mind: did the developers of AI systems intentionally design them to respond to prompts the way we do? Or is this merely a coincidence, a byproduct of reverse-engineering how intelligence might function?

It’s uncanny how similar this process seems. For example, large language models like ChatGPT or image diffusion models mimic this prompt-response dynamic, taking random noise or sparse data and turning it into meaningful output. Did this happen because the systems were inspired by how humans process information, or does this approach simply emerge as the most efficient way to deal with uncertainty and chaos?

Looking Forward

What excites me most is the idea that understanding this shared mechanism—between our brains and artificial systems—might lead to practical applications. If we can “prompt-engineer” ourselves just as effectively as we do AI, we could unlock entirely new ways to navigate the complexity of life.

Understanding this mechanism could change how we approach personal development, learning, and even therapeutic practices. Just as we’ve learned to craft effective prompts for AI systems, we might learn to better “prompt” ourselves for success, creativity, and resilience in our daily lives.

The future might not just be about engineering better AI prompts, but about understanding and optimizing the prompts that shape our own human experience.