Does the Adult Human Brain Make New Neurons? Part Two: Reading the Case Against

Abstract

The first part of this series examined the defence of adult human neurogenesis mounted by eighteen researchers in 2018. This part examines the reply, written by the group whose negative finding prompted the exchange. It is short, pointed, and argues that the defence made the wrong argument.

Their case has three components. That the field has attended to false negatives while ignoring false positives. That tissue quality is not a hypothesis to be raised in the abstract but a claim that can be tested, and that they tested it. And that a cell is not a young neuron merely because it carries a protein associated with young neurons — it must also look like one.

We find this reply harder to dismiss than the debate's public framing suggests, and we also think it stops short of establishing what it needs to establish. Both of those judgements are set out below.

1.A Reply, Not a New Experiment

The first thing to understand about this document is what it is. It is a letter in Cell Stem Cell, written in response to the commentary discussed in Part One, by Paredes, Sorrells, Alvarez-Buylla and colleagues. It contains no new dataset gathered to settle the dispute. It re-argues the interpretation of material already published.

That format has consequences for how much weight it can bear, and we will return to them. It also has an advantage: with no new results to present, the letter is unusually explicit about the reasoning, which makes it a better teaching document than most primary papers.

2.The Central Objection: Nobody Is Auditing the False Positives

The core complaint is directed at the structure of the defence rather than at any single point within it. Part One's argument was, in essence, a catalogue of ways a real signal might have been lost — degraded protein, masked antigen, stress before death. The reply's response is that this catalogue runs in only one direction.

Every method has two failure modes. It can miss something that is present, and it can report something that is not. The reply argues that the defence examines the first exhaustively and the second not at all — and that the methods relied upon most heavily for the positive case, the halogenated thymidine analogues and carbon-14 birthdating, are themselves susceptible to false positives and to artefacts of sample processing.

This is a fair structural point. A defence that explains away every negative result while treating positive results as self-validating is not a defence, it is an asymmetry. Whether the specific charge sticks against isotopic dating is a separate question, and here the reply is thinner than it needs to be: it raises the possibility without demonstrating it.

3.Tissue Quality as a Claim That Can Be Tested

The most substantive part of the reply concerns tissue. If the objection is that the material was too degraded to show what was there, then the answer is to work with material that cannot be dismissed on those grounds — and to include an internal control that proves the point.

They describe adult cases with an interval to fixation of under five hours, perfused with paraformaldehyde before the brain was removed. Perfusion fixation, delivered through the vasculature, preserves tissue far better than immersing a removed brain in fixative, and five hours is at the demanding end of what human postmortem work achieves.

The control is the more elegant move. In the same samples in which they found no young neurons in the hippocampus, they did find young neurons in the wall of the lateral ventricle — the other classical site of adult neurogenesis. The argument follows directly. If the antibodies worked, and the antigens survived, and immature cells were detectable a few millimetres away in the same block of tissue, then the absence in the hippocampus cannot be attributed to poor preservation.

This is, in our reading, the strongest single argument in the exchange. It converts a general anxiety about tissue quality into a specific, checkable claim, and it satisfies exactly the demand Part One made: show that the method would have seen the thing if the thing were there.

4.What a Young Neuron Should Look Like

The second substantive argument concerns identification, and it is where the two camps' standards visibly diverge. The reply's position is that a marker is necessary but not sufficient. A young neuron has a shape.

In children's tissue, they report, cells co-labelled for doublecortin and the accompanying markers were small — on the order of five to ten micrometres — elongated, and carried a leading process, the extension a migrating cell pushes ahead of itself. That is the morphology of a neuron on its way somewhere.

Their criticism of the opposing study is that the cells shown as evidence do not have this shape. They lack elongated nuclei and leading processes, and they are substantially larger — cells of twenty to thirty micrometres, against an expected five to ten. A cell can express a protein for reasons other than being what that protein usually indicates, and in tissue this heterogeneous, they argue, shape is the check on that.

The counter-consideration is one the reply itself supplies, though it does not treat it as damaging. They note that in children, doublecortin appears in neurons at several stages of maturation, including cells that have already grown recognisable dendrites and axons. If the marker spans a range of morphologies during development, then a strict morphological filter risks excluding real cells at the later end of that range.

5.The Problem With Counting Atoms

The carbon-14 method described in Part One is the positive case's most independent evidence, and the reply engages it directly, though briefly. Their observation is that within the isotopic dataset itself, many samples carry levels consistent with no addition of new neurons at all, and that whether this reflects real variation between people is unclear.

They extend the point into a general one about the positive literature: it is not internally consistent. Some studies report a steep decline with a negligible adult contribution. Others report substantial continuing neurogenesis into old age with very high variability between individuals. These are different claims, and the reply's argument is that treating them collectively as a single converging body of support obscures how much they disagree with one another.

6.Where Both Sides Actually Agree

It is easy to lose this in the framing of a dispute, and the reply is careful to state it. Both camps agree that new neurons continue to be recruited into the hippocampus of children. Both agree that disruption of that process could have lifelong consequences. Both regard postnatal neuronal recruitment as a genuine and important phenomenon.

The disagreement is narrower than headlines suggest: not whether the human hippocampus ever adds neurons, but whether it continues to do so at a meaningful rate in adults. That is a real question with real stakes, and it is a smaller one than 'does the adult brain regenerate'.

The reply also raises an open question that cuts across both positions and echoes a concession from Part One: how long does a human granule cell take to fully mature? Their answer is that it could be months and possibly years. If so, the entire framework of markers calibrated on a rodent timescale is measuring something whose duration in humans nobody has established.

7.What Would Settle It

The reply closes constructively rather than triumphantly, which is worth noting given the temperature of the exchange. Two proposals are made.

The first is that the field needs unified criteria for identifying a cell as a young neuron — an agreed standard for what combination of markers and morphology qualifies, so that two laboratories examining the same section would classify it the same way. Without that, the two camps are not disagreeing about the brain; they are disagreeing about definitions while believing they disagree about the brain.

The second is technical and, in the years since, has begun to happen. Single-cell transcriptomic profiling could identify the gene expression signatures that define each stage of maturation, replacing a handful of proteins chosen decades ago with a description of what the cell is actually doing.

8.Reading Two Papers That Cannot Both Be Right

Having read both documents closely, our assessment is that the reply wins on the specific question of tissue quality and loses on the general question of what follows from that. Perfusion-fixed material with a functioning internal control is a serious answer to the preservation objection. But demonstrating that one set of samples was adequate does not establish that the phenomenon is absent in general — particularly given the individual variability that both sides acknowledge.

What neither document can do, and neither pretends to, is resolve the question by argument. Both are commentaries. The dispute they describe is empirical, and it will be settled by methods that do not yet exist in mature form — which is the honest summary of where this stands.

There is a further move available, and it is the subject of Part Three. Both of these papers assume the question is about the hippocampus. It may be that the hippocampus is simply the wrong place to have been looking so exclusively.

Editorial Comment

We have presented this reply as forcefully as we presented the opposing case, because we think it is the stronger of the two on its central methodological point, and because a series that argued only one side would not be worth publishing. MindHeaven® makes no claim that any product influences neurogenesis in the human brain. Nothing in this exchange would support such a claim, and the researchers on both sides would be the first to say so.