A per diem rate set in 1986 for primate housing still shows up in neuroscience grant budgets, even though the circuit-mapping findings it was designed to support have mostly failed to replicate. The number survives because auditors reward consistency. The science does not survive because replication is slow, expensive, and rarely funded. This piece explains how that gap opened, why it persists, and how to read a grant budget with a skeptic's eye.

The Ledger That Outlived The Lab

Institutional per diem rates are set once and revised rarely. A rate approved in 1986 for primate housing might have been reasonable then, covering cage space, veterinary time, and feed. Today the same line item appears in budgets at a figure that no longer matches the actual cost of animal care. Grant officers copy it forward because it is defensible. Nobody asks whether the finding it supports still holds.

The neuroscience findings from that era often did not travel. Circuit-mapping claims from small primate studies, built on bespoke lab conditions and tiny sample sizes, were cited for decades. When replication attempts came, many were underpowered or used different protocols. Some claims quietly disappeared from review articles. The per diem did not. It became an administrative artifact, a number with no scientific counterpart.

This is the tension at the center of research economics. Accounting memory is durable because auditors demand it. Scientific memory is fragile because it depends on replication infrastructure that funding cycles rarely support. A lab can lose its finding and keep its rate. The rate may outlive the lab itself.

How A Rate Became A Fixture

Per diems calcify through a simple mechanism. A grant officer inherits a budget template from a predecessor. The template lists animal costs at the 1986 rate. The officer submits the budget. The auditor sees continuity and approves. No one flags the rate because no one is paid to flag it. The incentive runs the other way: consistency looks like competence.

Neuroscience animal costs are often bundled into overhead, which hides the mismatch. A university negotiates an indirect cost rate with federal funders, and that rate covers facilities and administration. The per diem sits inside that bundle. Principal investigators see a total, not a line item. The stale rate becomes invisible until a young lab tries to build a start-up budget from scratch.

Reasonable people disagree about whether this is a bug. Some administrators argue that stable rates protect long-term planning. A lab that knows its animal costs can budget three years out. Critics counter that stability is only useful if the number is accurate. A stable wrong number is a stable misallocation. Both sides agree the revision process is slow.

The Findings That Did Not Travel

The primate circuit-mapping work of the 1980s was ambitious. Researchers traced connections between cortical areas using tracer injections, producing wiring diagrams that shaped textbooks. The studies were technically demanding and often relied on a handful of animals. That small n was not a flaw in itself. It became a problem when later researchers treated the diagrams as settled maps rather than provisional sketches.

Replication attempts in the 1990s and 2000s often used different tracers, species, or cortical parcellation schemes. Results diverged. Some connections held. Others did not. A few became contested. The literature accumulated citations faster than confirmations, a pattern that later became familiar across neuroscience. Publication pressure favored novel claims over confirmatory work, so the corrections arrived slowly and quietly.

Not everyone reads the divergence as failed findings. Some tract-tracing specialists argue that tracer sensitivity and cortical parcellation schemes vary enough that apparent contradictions reflect methodological differences rather than errors. For instance, a 2003 replication attempt by Smith and colleagues reported that only a subset of the original connections could be confirmed using modern tracers, but the authors cautioned that negative results in tract tracing are notoriously hard to interpret. The debate remains open, which is precisely why the per diem's persistence is so striking: the accounting never registered the uncertainty.

Why Numbers Outrun Conclusions

Overhead rates calcify faster than theories because they are negotiated on multi-year cycles and defended by institutions. A theory can be revised in a single paper. An indirect cost rate might take a decade to renegotiate. Funding cycles are shorter than revision cycles, which means a lab may receive three grants under a rate that no longer reflects reality before anyone revisits it.

Some funders have tried to address this by requiring cost basis documentation. Others have moved to modular budgets that reduce line-item scrutiny. Neither approach solves the underlying problem. A modular budget hides the stale rate rather than fixing it. Cost basis documentation helps only if someone reads it and acts.

What This Costs A Young Lab

Start-up budgets inherit legacy per diems. A new principal investigator building a primate or rodent program may find that animal-care costs consume a larger share of the budget than expected, because the inherited rate understates true costs. The shortfall has to come from somewhere. Often it comes from imaging time, which is flexible and easy to cut.

Indirect cost recovery varies by institution, which means two labs doing similar work can face very different effective costs. A lab at a well-endowed private university may absorb the mismatch through endowment. A lab at a public institution may not. Early-career researchers absorb the difference, often without realizing the rate they inherited was set before they were born.

The pattern echoes other infrastructure legacies. A 1973 rat colony kept one memory paper alive through four funding droughts, long after the original claims were questioned. Deep-sea mooring fees outlast the paleoclimate record they were built to recover. The mechanism is the same: infrastructure persists, findings do not.

Reading A Grant Budget Skeptically

When you encounter a per diem line item, ask when it was last justified. A rate with no recent cost basis is a hypothesis, not a fact. Request the institutional cost study that produced it. If none exists, treat the number as provisional.

Separate legacy line items from active needs. A budget that lists animal costs at a 1986 rate alongside 2024 imaging costs is mixing eras. Flag the mismatch before submission, not after the first progress report.

Check whether cited findings have replicated. A budget that justifies animal work by citing a 1987 circuit-mapping paper should be read alongside later replication attempts. If the finding is contested, the cost basis is contested too.

Request institutional cost basis in writing. Verbal assurances that the rate is standard do not survive an audit. A written cost study does. If the institution cannot produce one, that absence is information.

Treat budget continuity as a hypothesis. A number that has appeared in ten consecutive grants is not necessarily correct. It may simply be unexamined. Continuity is a claim about the past, not evidence about the present.