Someone told me they remembered a trick I had done at a corporate event in Graz. They described it in detail. The description was confident, specific, and enthusiastic.
The only problem was that I had not done that trick. Not at that event, and not in the way they described. What they remembered was an amalgam of two different effects from two different parts of the evening, combined with what I can only assume was their brain’s editorial intervention to make the combined memory more coherent and more dramatic than either original.
They were not lying. They were not misremembering in any deliberate way. They were doing what all human brains do: they were reconstructing.
Memory Is Not Recording
Daniel Gilbert’s “Stumbling on Happiness” contains one of the most useful things I have read for thinking about what magic performance actually does to an audience. His central argument about memory is that it works nothing like a recording device. The brain does not store complete records of experiences and then retrieve those records intact. Instead, it stores fragments, impressions, emotional markers — and when recall is needed, it reconstructs a full memory from those fragments, filling in the gaps with what seems plausible given everything else it knows.
The filling-in happens automatically and invisibly. The person doing the remembering has no awareness that reconstruction is occurring. The memory feels like a recording even though it is a synthesis.
This has enormous consequences for how we think about perception and experience generally. But it has very specific consequences for magic.
The Audience Edits While They Sleep
Here is what I have observed, and what Gilbert’s framework explains: audiences tend to remember effects as slightly cleaner, slightly more impossible, and slightly more dramatic than they actually were.
Not wildly different. But measurably improved. The effect they remember having watched is often a better version of the effect they actually watched.
The brain, during consolidation and recall, tends to smooth out rough edges. The hesitation before the reveal gets compressed or eliminated in memory. The moment where you had to reset something and the audience was briefly uncertain about what was happening gets dropped. The pacing feels tighter. The impossibility feels starker.
I have tested this informally across many performances. When I follow up with people who saw a show, and I ask them to describe specific effects, the descriptions are almost always slightly elevated above what I know the reality to have been. The memory has been curated.
This is not a quirk. This is the normal operation of human memory. Gilbert’s research makes clear that this happens consistently and predictably. Memory is not neutral storage — it is active interpretation, and the interpretation tends toward coherence and drama.
What This Means for Effect Design
If you know that memory tends to improve upon reality, the design implication is significant: the single most important moment in an effect is the final image. Not the setup. Not the method. Not the arc of the effect as it develops. The final image.
Because that is what the memory will anchor to, and everything else in the effect will be reconstructed in light of that anchor.
An effect with an extraordinary final image and a merely competent build will be remembered better than an effect with a merely competent final image and an extraordinary build. The brain stores the peak emotional moment and reconstructs backward from it.
This completely changed how I think about endings. I used to think about endings as the resolution — the payoff after the structure of the effect has done its work. That is not wrong, but it is incomplete. The ending is also the anchor for memory. The ending is what the audience will remember and what they will reconstruct the rest of the experience in relation to.
The final image needs to be extraordinary. Everything else can be merely excellent.
The Peak-End Rule in Performance
Gilbert’s framework connects to research by Daniel Kahneman — who appears in several places in my reading history — on what is sometimes called the peak-end rule. People judge experiences primarily by their most intense moment and their final moment, rather than by an average of the full experience. The duration of the experience matters much less than these two data points.
A show that has a transcendent peak and a powerful ending will be remembered more positively than a show that was consistently very good but ended weakly, even if the second show was objectively higher quality across its duration.
This sounds like it argues for neglecting the middle of a performance. It does not. The middle builds the emotional bank account that the peak and ending draw from. But it does argue for prioritizing design energy on the moments that memory is most likely to retain.
The peak, whenever it occurs, needs to be genuinely remarkable. The ending needs to be clean, strong, and emotionally resonant. These are the two moments the brain is most likely to store and anchor its reconstruction to.
The Reconstructive Bonus
Here is the aspect I find most interesting, and most useful for thinking about why magic seems to get better in people’s memories rather than worse.
When the brain reconstructs a memory of something it found surprising and inexplicable, it tends to make it more surprising and more inexplicable. The gaps it fills in tend toward the more dramatic option rather than the more mundane one.
This makes sense from a cognitive efficiency perspective. If something generated a strong emotional response — the surprise and disorientation of a well-executed magic effect — then the brain’s reconstruction of that thing should probably also generate a strong response. So it fills the gaps in ways that maintain or increase the emotional intensity.
What this means in practice: a magic effect that genuinely astonishes someone will tend to grow in the memory. They will not rationalize it downward. They will, if anything, rationalize it upward — remembering it as more impossible, more clean, more extraordinary than it was.
This is the opposite of what happens with things that leave the audience in a state of analytical skepticism. When someone watches an effect and thinks “I almost figured out how that works,” the memory tends to degrade. Their brain fills the gaps with the possible explanation they were reaching for. The effect shrinks in memory rather than growing.
The Performance I Did Not Do
The person in Graz who described an effect I had not performed — I found this clarifying rather than troubling. They were not confused about whether something wonderful had happened. They were confident something wonderful had happened. They had simply stitched together the details from their reconstruction of the evening rather than from accurate storage of specific events.
What this told me was that the emotional impact of the evening was strong enough that their brain had flagged it as worth preserving and worth enhancing. The reconstruction served the emotional truth of the experience, which was: this was remarkable.
That is, in a strange way, a success. The memory they have is more positive than the reality was. The experience improved in the having of it.
The Danger Side
There is a danger side to this that is worth acknowledging. If the audience’s memory is constructive rather than reproductive, it also means that negative emotional markers get reconstructed in the same way. A moment of awkwardness or confusion that generated discomfort will tend to be filled in during reconstruction with more confusion and more awkwardness. The memory of something that did not land can be worse than the reality.
This is why I think about the emotional markers an audience is likely to carry away from a performance, not just the specific effects. If the dominant emotional markers are confusion, discomfort, or skepticism — even if those are only present in small patches — the reconstruction will tend to fill gaps in that direction.
You want the emotional markers the audience carries to be the ones you chose. Astonishment. Warmth. Delight. Productive unease. If those are the markers, the reconstruction will serve them.
Design for the Memory, Not Just the Moment
The practical upshot of all of this is a shift in how I evaluate effects after performing them. I used to ask: did that land? Meaning: did I see the response I was hoping for in the moment?
I now also ask: what memory are they going to reconstruct? What emotional markers am I leaving them with? What is the final image they are going to anchor their reconstruction to?
These are design questions as much as performance questions. They affect how I structure effects, where I place the emotional peak, and what happens in the final five seconds before the effect is over.
Gilbert would probably not have expected his research on memory and happiness to end up informing the design of mentalism routines by an Austrian consultant. But the principle travels.
Memory is not a recording. It is an interpretation. Design your performances knowing that the audience’s interpretation will do some of the work for you — if you give it the right material to work with.