Decision fatigue in photography: what the research actually says

Every article about culling eventually reaches for decision fatigue, usually with a line about judges and parole hearings. It is a tidy story and the science behind it is far less settled than the retelling suggests. This is an honest look at what the studies found, what failed to reproduce, and what a photographer should reasonably do with any of it.

The story you have heard

The popular version goes like this. Willpower works like a tank. Every decision you make draws it down, and once it is low your later decisions get lazier, more impulsive, or default to whatever takes the least effort. Hence the 1 AM cull where you keep everything, or clear whole scenes on the thumbnail.

The research tradition behind it is real. Roy Baumeister and colleagues published a long run of experiments from the late 1990s onward on what they called ego depletion: participants who exerted self-control on one task appeared to do worse on an unrelated task afterward. It became one of the most cited ideas in social psychology, and it is where nearly every business article about decision fatigue ultimately traces back to.

Here is the part that rarely makes it into the retelling. Over the last decade, that literature has been through a hard reckoning, and researchers still disagree about whether the core effect exists. If you are going to reorganize your working week around an idea, you should know how firm the ground is.

What happened when people tried to reproduce it

In 2016, a group of researchers ran a preregistered replication across 23 laboratories with 2,141 participants, published in Perspectives on Psychological Science. Preregistered means the analysis was locked in before the data came in, which closes off a lot of accidental massaging. The multilab replication reported an effect size of d = 0.04 with a confidence interval that included zero. In plain terms: if there is an effect there, it is close to nothing.

More recently, researchers looked for decision fatigue in real working data rather than in a lab. A 2025 paper in Communications Psychology analyzed 231,076 phone calls handled by 174 specialized nurses at a Swedish national telephone triage service. That is about as close to a high-volume sequential decision job as you can get. The authors reported no evidence for decision fatigue, with results consistently supporting the null, and wrote that their findings cast serious doubt on decision fatigue as a general effect for sequential decisions.

Two honest caveats, because this cuts both ways. A failed replication is not proof that nothing is there. And triage nurses following clinical protocols are not photographers comparing ten frames of the same laugh. What these results do establish is that the confident version you read in productivity blogs is not something the evidence currently supports, and that serious researchers disagree about it right now.

The related findings are messier too

Decision fatigue rarely travels alone. Two other ideas get cited alongside it in culling articles, and they deserve the same scrutiny.

Choice overload

The famous one is Iyengar and Lepper's 2000 jam study, where a tasting table with 24 varieties drew a bigger crowd than one with 6, but shoppers at the smaller table were far more likely to actually buy. It is a lovely result and it maps neatly onto a folder of 3,000 frames. But a 2010 meta-analysis in the Journal of Consumer Research pooled 63 conditions from 50 experiments with 5,036 participants and found a mean effect size of virtually zero, with no reliable way to predict when choice overload shows up and when it does not. Individual studies vary a lot. The headline does not hold as a general law.

Comparison versus absolute judgment

This one is on firmer ground, mostly because it is a broad statement about perception rather than a single striking experiment. People are generally better at telling which of two things is bigger, sharper or brighter than at scoring one thing on its own, and the smaller the difference, the harder either task gets. That is why a burst of ten nearly identical frames is genuinely difficult, and why rating every frame from one to five asks your visual system for its weakest operation thousands of times over.

We leaned on some of this framing when we wrote about why picking the right photo is so hard. The perceptual part of that argument stands up well. The willpower part deserves the hedges on this page, and it is worth saying so plainly rather than quietly.

What none of this says about photographers

The most important sentence in this article is the least satisfying one. Not one of the studies above looked at a photographer culling a shoot. There is no published research we could find on selection quality across a long culling session, no keep-rate data by hour, nothing. Anyone who tells you science proves photographers suffer decision fatigue is filling a gap with confidence.

Two more limits worth naming. Nothing here is a health claim, and we are not qualified to make one: if you feel genuinely unwell after long sessions, that is a conversation with a professional, not something a photography blog should be answering. And an effect that is small or absent on average in a lab can still be a real pattern for a specific person doing a specific job. Averages hide individuals in both directions.

Why the practical advice survives the argument

You could read all of the above and conclude that culling in blocks is superstition. That would be the wrong lesson, because the workflow advice never depended on the willpower mechanism in the first place. It rests on things that are much cheaper to believe.

• Fine visual comparison is demanding work. You do not need a psychology paper to know that three hours of judging near-identical frames on a bright screen is harder than one hour.
• Switching between kinds of thinking has a cost. Checking focus and judging a moment are different jobs. Interleaving them thousands of times is slower than batching them, and you can time that yourself.
• The interventions are free and reversible. Working in 45-minute blocks, taking real breaks, and culling in daylight cost nothing. You do not need a confirmed mechanism to justify a change that has no downside.
• What you are protecting is the delivery. The point was never willpower. It is whether the gallery a client receives was picked with the same standard from the first frame to the last.

The mechanics of all of that, session length, breaks, screen and room setup, and the hour you start, are laid out properly in why culling is exhausting. That is the practical companion to this page, and it is the one to send to someone who does not care what a meta-analysis is.

Run the experiment on your own shoots

The honest replacement for citing psychology at yourself is collecting your own evidence. It takes about three shoots and a notes app.

1. Log the blocks. Start time, stop time, how many frames you got through, how many you picked. That gives you a keep rate per block, which is the number that actually moves.
2. Split a similar shoot two ways. Cull one family session first thing in the morning and a comparable one late at night. Compare the keep rates and the time per hundred frames.
3. Audit your own tail. Take the last block of a late session and review it again the next morning. Count how many calls you would change. That count is the whole argument, in your own handwriting.
4. Write the criteria down first. If your standard is a feeling rather than a list, your keep rate will drift for reasons that have nothing to do with fatigue. Start with building your culling criteria so the measurement means something.

Be fair about what this is. Your own log is not a controlled study either, you know what you are testing, and expecting to find an effect makes you likelier to find one. But it is evidence about you, doing your actual job, which beats a lab result about undergraduates squeezing a handgrip. For a sense of what normal looks like before you start, we collected some reference figures in culling time benchmarks and in how long culling actually takes.

Where software honestly fits

The fatigue argument sells culling software, and we sell culling software, so this section should be read with that in mind. Here is the version we can defend.

An AI first pass does not make you a better judge and it does not top up any tank. What it does is reduce the number of decisions you personally make, and it takes the most repetitive ones first: focus, closed eyes, near-identical frames from a burst. A machine is genuinely indifferent at frame 3,000 in a way no person is, and that is a fair claim because it is a statement about software, not about your brain.

What is left is still work. Reviewing a proposed selection is a review job, and it has its own kind of tiredness, which is why the honest promise is fewer decisions rather than none. We went through what these models get right and wrong in can you trust AI culling, and the answer is neither yes nor no.

FAQ