Cumulative level test: Beyond the point neuron: circuit mechanisms beneath behaviour and the methods that can test them

Check understanding, separately from reading

12 questions · 80% to pass · no timer. This longer test revisits both units in the level. These authored questions assess recognition and application of the taught distinctions, not professional qualification. You may review the lessons and retry. Repeat attempts reuse the question bank; a remembered answer is not proof of transfer to a new situation.

Saving a lesson records reading only. Previous lesson completions have not been converted into passing scores. You can continue reading without a pass; the assessment remains unpassed.

Question 1

A paper claims that a class of neurons performs linear integration, based on a regression of somatic voltage on the number of stimulated spines with a slope of 0.9 millivolts per spine and an intercept of 2 millivolts, fitted across 12 stimulation intensities. You have one extra recording session. Which single manipulation would most directly test the linearity claim, and what would the result look like under a nonlinear mechanism?Hold the number of stimulated spines constant and move them between one branch and several branches: a linear integrator should produce the same somatic voltage for the same count regardless of arrangement, whereas a supralinear dendritic mechanism should produce a larger response when the inputs share a branch, and blockade of the voltage-dependent conductance should reduce the arranged condition toward the split condition.Fit the same regression separately for the proximal and distal halves of the dendritic tree and compare the slopes, because differences in slope between locations are the definition of nonlinear integration.

Question 2

Two reviews of the same astrocyte case make opposite recommendations: one concludes that astrocytes are integral synaptic elements, the other that the key release evidence is unreliable. A researcher must decide whether to spend a year on a follow-up experiment. Using the two positions, what design would produce a result that both reviews would have to accept?A larger experiment using the same dnSNARE construct in more animals, because the criticism concerned statistical power rather than the validity of the construct.A within-preparation design in which the release machinery is disrupted only in astrocytes using a targeting construct whose leak into neurons is quantified, with the receptor on the neuronal side manipulated independently and the synaptic outcome measured in the same animals, so that the result does not depend on the specificity of a single genetic line.

Question 3

You are editing a textbook chapter that currently states that critical periods end when perineuronal nets form, and that adult brains therefore cannot change with experience. Two sentences need rewriting. Which pair of sentences is defensible?Matrix maturation around fast-spiking interneurons is one contributor to the decline of plasticity, alongside changes in inhibition and other factors; and adult experience-dependent change is graded rather than absent, as shown by small reversible ocular dominance shifts after a few hours of monocular deprivation in adults, with larger effects requiring much larger doses.Matrix maturation closes critical periods, which is why digesting chondroitin sulfate chains reactivates plasticity; and adult brains retain a latent capacity for change that is released whenever the matrix is removed.

Question 4

A funding panel receives two proposals about adult human hippocampal neurogenesis. Proposal one will count doublecortin-positive cells in autopsy tissue from 40 donors aged 20 to 80 using one antibody and existing protocols. Proposal two will process surgical and autopsy tissue from the same donors through two protocols, use two immaturity markers plus morphology, and pre-register the counting rules. Which should the panel prefer, and on what reasoning?Proposal one, because a larger donor sample across a wide age range gives a better estimate of the population quantity and will settle the disagreement by accumulating evidence.Proposal two, because the dispute in this literature turns on identification, processing and quantification rather than on sample size, so a one-marker count under a single protocol would produce another number that cannot be compared with the existing discordant results.

Question 5

A pharmaceutical company proposes to develop a compound that enhances microglial synaptic pruning during adolescence, on the grounds that microglia prune synapses in development and that altered pruning is implicated in psychiatric disorders. As a scientific adviser, what is your primary objection and what evidence would be required to overcome it?The proposal inverts an unfinished inference chain: engulfment and elimination have been demonstrated in the mouse retinogeniculate system with circuit-specific and sometimes negative results, no measure of synapse elimination exists in living human brain, and disease involvement is presented as an implication of the model rather than an established mechanism, so human efficacy and even the direction of benefit are unsupported.The proposal is premature only because the relevant pruning window in humans has not been dated precisely, so the compound should be tested once the adolescent window is defined by imaging studies.

Question 6

A preprint reports that a model's middle layer best matches a cortical area, with a correlation of 0.5 that was estimated after averaging 200 trials per participant across 15 participants, and it argues that the correspondence explains the area's function. Meanwhile a recording paper reports that only 40 per cent of its sorted units met an isolation threshold. How do the two reporting problems belong to the same family?They belong to the same family because both reflect insufficient sample size, and both would be resolved by collecting more data: more participants in the modelling study and more recording sessions in the electrophysiology study.Both are quantification problems in which a summary number is presented without the quantity that bounds it: the model result needs held-out generalisation, an account of how the layer was selected and the precision of a correlation estimated across participants, while the recording result needs the discarded fraction and a population-level error estimate, because both a correlation of 0.5 and a unit count are inferences whose uncertainty is not in the reported figure.

Question 7

A study reports that a polygenic score for a behavioural trait correlates with a brain imaging measure at r equal to 0.15 in 8,000 participants, that the imaging measure correlates with the behaviour at r equal to 0.2, and concludes that the imaging measure mediates the genetic influence. Which elements of this argument are quantitatively and conceptually weak?Both correlations correspond to 2.25 and 4 per cent of variance, so the measures are largely independent of the predictors and of each other, a mediation claim requires the product of the paths and a test of the indirect effect, and the imaging measure's reliability attenuates the observed associations so the true genetic effect may exceed what was detected.The argument is weak only because 8,000 participants are too few for a genetic study, given that complete discovery of the architecture of brain traits requires sample sizes in the millions.

Question 8

You have one month and a fixed budget to test whether a specific cell population is necessary for a behaviour. Choose between a chemogenetic suppression study with 30 animals per group and an optogenetic suppression study with 15 animals per group performed within subjects, and justify the choice with an explicit account of the artifacts each design must bound.Prefer the chemogenetic design, because it avoids the heating problem entirely and therefore requires fewer controls, making it the cheaper way to establish necessity.Prefer the within-subject optogenetic design if the behaviour can be measured repeatedly, because within-subject comparison removes between-animal variance and raises power at the same sample, at the cost of adding sham illumination, a temperature estimate and ideally an ion-swapped opsin; the chemogenetic design requires fewer technical controls per animal but needs additional groups for receptor-free and vehicle conditions and a dose-response, which will consume the 30 animals before the question is asked.

Question 9

Two years from now, a laboratory publishes a study of adult human hippocampal neurogenesis using tissue from the same donors analysed through both a fresh frozen protocol and a standard fixed protocol, with two immaturity markers, morphology criteria and pre-registered counting rules. The result is that the fixed protocol detects almost no young neurons while the fresh frozen protocol detects a small number. What is the correct interpretation, and what would remain unresolved?The result identifies processing as a sufficient explanation for at least part of the historical discordance, since the difference appears within the same donors, and it leaves the functional question unresolved: whether the detected cells are functionally integrated young neurons, whether their number has any behavioural consequence, and whether measures from different protocols can be pooled.The result settles the debate in favour of the positive studies, because cells detected under conditions that preserve messenger RNA are real while cells lost to fixation were never there to be counted.

Question 10

A theory paper argues that a specific dendritic nonlinearity implements a Bayesian inference scheme, showing that a model with this nonlinearity reproduces three published sets of findings. Which additions would turn this into a constraining claim rather than a description, and how would you test them?The model must be fitted to more findings, across more brain regions and more species, until the number of reproduced results is large enough that the account can be treated as established.The model must make a prediction about a measurement not used in its construction, the prediction must be tested at a stated generalisation level in held-out data, and the comparison must be against a rival model that fits the existing findings equally well but predicts differently, with the discriminating experiment then run using a manipulation that separates the two mechanisms.

Question 11

Compare two mechanistic claims: that a dendritic plateau potential depolarises the soma enough to influence spike output, and that an astrocyte calcium microdomain triggers gliotransmitter release that changes a synapse. Both rest on measured signals and both involve a step from signal to effect. What distinguishes their evidential standing, and what does each need to be comparable?The dendritic case has a measured downstream consequence in the same preparation, namely a sustained somatic depolarisation of 10 to 20 millivolts from a local event, whereas the glial case requires an unevidenced intermediate release step whose most-used evidence was challenged by a documented tool-specificity defect, so the astrocyte claim needs an independent demonstration of release plus a neuronal-side manipulation, while the dendritic claim needs a demonstration that the depolarisation changes behaviour.The two are equivalent in standing because both are mechanistic claims about a measured signal producing a downstream effect, and in both cases the link would be established in the same way, namely by repeating the measurement in more preparations.

Question 12

A trainee proposes a study with these features: 12 animals per group, optogenetic activation of a projection at 25 milliwatts, no sham illumination group, chemogenetic suppression at 10 milligrams per kilogram of ligand with vehicle controls only, one anatomical marker for cell identity, and analysis of 14 behavioural outcomes with the significant ones reported. Reprioritise the study's resources and give the three changes that would most improve what it can conclude.Keep the design and add more anatomical markers for cell identity, since the credibility of a circuit claim rests on demonstrating that the manipulated cells are the intended population and the behavioural outcomes are secondary.Move resources from the number of outcomes to the controls and the sample: power the design for a stated minimum effect of interest, roughly 63 animals per group for a 0.5 standard deviation effect at 80 per cent power, add the sham and temperature controls plus a receptor-free ligand group, and pre-specify one primary outcome with the remaining outcomes labelled exploratory, because 14 uncorrected tests carry a family-wise error rate above 0.5 and the current design risks concluding nothing from any of them.

Continuing does not mark this assessment passed.

Progress stays in this browser, on this device, not an account or cloud backup. Clearing browser data removes it. Visits never count as study. Previous untimed legacy completions are not invented as study days.