In 2011, a study of roughly 1,100 parole hearings in Israeli courts produced one of the most quoted findings in modern legal psychology: prisoners who appeared before a judge early in the morning or right after a food break received favorable rulings about 65 percent of the time, while those unlucky enough to appear just before a break received almost none. The graph looked like a set of stairs. Favorable decisions dropped steadily through the morning, jumped back up after lunch, and dropped again through the afternoon. The judges, presumably, believed each case was decided on its merits.
That single paper launched a decade of arguments about whether justice really does depend on when a judge last ate.
The answer, on the current evidence, is more complicated than the headline. But the underlying question, whether ordinary biological states quietly shape decisions that reshape lives, has held up better than any single explanation for it.
What the original study actually claimed
The 2011 paper by Shai Danziger, Jonathan Levav, and Liora Avnaim-Pesso, published in the Proceedings of the National Academy of Sciences, tracked eight Israeli parole judges over ten months. The authors found that the probability of a favorable ruling was highest at the start of each of three daily decision sessions and declined sharply as each session wore on, before rebounding after the judge took a break. The most widely circulated framing came from a National Geographic write-up at the time: justice, in this dataset, was quite literally served after lunch.
The proposed mechanism was the mental depletion that occurs when making a long series of consequential choices. Judges under this kind of strain, the theory went, become mentally depleted, and a depleted decision-maker defaults to the safer, more conservative option. In a parole hearing, denying release is the low-risk default. A hungry, tired judge, on this reading, does not become crueler. Just more cautious.
This is one study, not settled consensus. Later analyses have argued the original effect was partly an artifact of case ordering: represented prisoners often went first in each session, and unrepresented prisoners tended to cluster at the end. That critique matters. But it did not close the question, because the broader phenomenon it pointed to, that human decision-makers drift as the day drags on, has been observed across many domains that have nothing to do with parole.
Time of day as a quiet variable
Fatigue and time-of-day effects appear in medicine, in aviation, in education, and in everyday professional judgment. Patterns have emerged showing doctors prescribing more antibiotics later in the day, radiologists missing more findings after extended reading sessions, and teachers grading the same essay differently depending on when in the day they read it.
A 2022 analysis in Frontiers in Sociology found that academic assessments shift measurably across the hours of a working day, with the direction and size of the drift depending on the type of task and the assessor’s own daily rhythm. The finding is modest on its own. Placed next to the medical, aviation, and legal literature, it becomes harder to argue that judges are somehow immune to a pattern documented almost everywhere else humans make sequential decisions.
A broader review of cognitive biases across four occupational fields, including legal work, catalogs how cognitive shortcuts—such as relying heavily on initial information or the ease with which examples come to mind—along with depletion effects shape the judgments of trained professionals. Training and expertise dampen some biases. They do not eliminate them. The reviewers were careful to note that mental depletion from sustained decision-making is real in the aggregate even where its size in any single study is contested.
That is roughly where the honest state of the evidence sits. Judges are not robots. Nor is every unfavorable afternoon ruling the fault of a missed sandwich.
Why the hungry-judge story spread anyway
The reason the finding traveled so fast is not really about food. It is about the discomfort of the alternative.
Courts present themselves as engines of considered reason. The robe, the raised bench, the formal address, the written opinion, all of it signals that what happens inside is the opposite of impulse. When a study suggests that something as trivial as a glucose dip can nudge a life-altering decision, the appeal of the story is partly relief. A biological explanation feels more manageable than the harder possibilities: entrenched bias, structural inequity, or the sheer randomness of which judge a defendant draws.
Islamic jurisprudence noticed something like this a very long time ago. Classical scholars warned judges against ruling while hungry, angry, drowsy, or distracted, and treated these states as disqualifying rather than incidental. A discussion in Al Hakam traces this tradition and notes how closely it anticipates the modern behavioral literature. The intuition that internal state contaminates judgment is not new. What is new is the attempt to measure it.
Decision fatigue in the rest of the courthouse
Judges are the visible decision-makers, but they are far from the only ones making consequential calls under load. Prosecutors decide which charges to file, which pleas to accept, and which cases to drop. Public defenders triage caseloads that dwarf what any single lawyer can meaningfully attend to. Probation officers write pre-sentence reports that heavily shape the judge’s eventual number. Each of these roles produces long chains of sequential judgments, often on incomplete information, often late in the day.
A 2025 Frontiers study of judges and prosecutors examining sentencing disparity found that legal actors themselves were acutely aware of inconsistency in outcomes and offered a mix of explanations for it: caseload pressure, individual philosophy, the emotional weight of certain case types, and the accumulated wear of hearing dozens of matters back to back. The professionals did not describe themselves as machines. They described themselves as people trying to hold a stable standard across conditions that made stability difficult.
That framing matters. It moves the conversation away from the caricature of the grumpy hungry judge and toward the more accurate picture: a system that asks humans to perform consistently under conditions that reliably produce drift.
What the biological story does not explain
Any account of sentencing that stops at hunger and fatigue is incomplete in a way that matters.
Time-of-day effects are small compared with the disparities that show up when the data is cut by race, income, or geography. Research indexed by Nature on federal criminal sentencing has repeatedly documented differential treatment across judicial districts and demographic groups that no glucose-based theory can absorb. The same offense, prosecuted in different districts, produces meaningfully different average sentences. The same offense, committed by defendants of different racial backgrounds, produces meaningfully different average sentences within the same district.
California’s response to this problem is instructive. Under the state’s Racial Justice Act, defendants can raise claims that racial bias influenced their charge or sentence. Few such claims have succeeded, partly because the standard for proof is high and partly because the bias the law aims at is diffuse rather than episodic. The problem is not usually a single visible moment of prejudice. It is a pattern that shows up only when many cases are stacked together.
Mental depletion and demographic disparity are not competing explanations. They interact. A tired judge, defaulting to the conservative option, will lean on whatever priors are already in place. If those priors carry the weight of a lifetime of cultural exposure to who looks dangerous and who looks safe, the biological drift and the social bias push in the same direction.
Where the folk version goes wrong
The version of the hungry-judge story that circulates online usually skips three things worth keeping.
First, the effect size in the original study was startling precisely because it was so large. Most findings related to sustained decision-making are smaller, subtler, and easier to miss without careful statistical work. The parole graph is memorable partly because it is unusually clean, which is one of the reasons later researchers scrutinized it so closely.
Second, the parole context is not the same as a full criminal trial. Parole hearings are short, repetitive, and structured around a binary decision. Sentencing after a jury trial involves briefing, argument, written recommendations, and often days or weeks between the verdict and the sentence. The mechanisms that push a rushed parole ruling toward denial are not identical to the ones shaping a sentencing memo.
Third, the finding was about ordering within a session, not simply about the clock. A ruling at 11:30 a.m. was not doomed. A ruling at the tail end of a long uninterrupted block of cases was where the drift showed up. The intervention the study implied was not later lunches. It was more breaks.
Consistency as an institutional problem
One useful reframe: judicial inconsistency is not primarily a story about individual weakness. It is a story about system design.
If a court schedules thirty parole hearings in a single session without breaks, the schedule itself is a variable in the outcome. If a public defender inherits a caseload that guarantees rushed preparation, the caseload is a variable in the outcome. If sentencing guidelines leave wide discretion without structured checkpoints for reflection, that architecture is a variable in the outcome.
A recent examination of consistency in AI legal tools inadvertently sharpened this point. In one analysis of 12,000 tests of legal AI systems, the same question posed multiple times produced meaningfully different answers. The comparison with human judges is not that machines are worse or better. It is that consistency, in any complex judgment system, has to be engineered. It does not emerge on its own from good intentions or expertise.
Human judges have the advantage of moral responsibility, professional training, and the ability to explain a decision in language a defendant can understand. They have the disadvantage of being embodied, tired, distracted, and subject to the same drift the rest of the species carries into work.
What defense lawyers already know
Trial attorneys who handle a lot of hearings tend to develop an unspoken map of their local courthouse. Which judges rule sharper in the morning. Which prosecutors return calls before lunch and go silent after. Which chambers hear motions in blocks that leave the last case in a room of exhausted people. None of this appears in the rules of civil or criminal procedure. All of it factors into scheduling requests.
Defense lawyers who ask to reset a sentencing hearing to a morning slot, or who fight to be earlier on a crowded calendar, are working from the same intuition the parole research formalized. Whether the underlying effect is mental depletion from sustained decision-making, glucose depletion, or simply the accumulated weight of a hard morning, the operational lesson is the same. Sequence matters. Timing matters. The invisible variables around a case can move the outcome as much as some of the visible ones.
This is general information about how these situations tend to unfold, not legal advice. Laws and procedures vary by jurisdiction, and a licensed attorney is the right source for guidance on any specific case.
What the research supports, carefully stated
Pulling the strands together, the defensible reading of the current evidence looks something like this. Judges, like other professionals making long sequences of consequential decisions, show measurable drift over the course of a working session. The size of the drift in any single study is contested. The existence of the general pattern, across decades of research in medicine, education, and law, is not.
Hunger and fatigue are shorthand for a wider phenomenon: the human capacity for careful, individualized judgment is finite, and it depletes under load. Rested, fed, and unhurried decision-makers do not become perfectly fair. They become somewhat more likely to give each case the attention its facts deserve.
None of this displaces the larger drivers of sentencing disparity. Race, geography, prosecutorial discretion, and access to competent counsel remain the heavy variables in the data. The hungry-judge finding is not an alternative to those explanations. It is a reminder that even inside a single judge’s own courtroom, on a single day, the same case can catch two different versions of the same person.
The parole prisoners who happened to be scheduled at 9:15 in the morning did not know they had drawn a better hand than the ones scheduled at 4:45. The judges did not know it either. That is what makes the finding uncomfortable, and what makes it worth taking seriously even where the numbers keep getting re-examined.
Justice is supposed to be blind. The research keeps suggesting it is also, at various points in the day, tired.