Probing perceptual decisions in rodents

Probing perceptual decisions in rodents Carandini Churchland

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

Probing perceptual decisions in rodents 2013 Nature neuroscience 10.1038/nn.3410 Carandini2013-probing-perceptual-decisions-in-rodents Carandini2013-probing-perceptual-decisions-in-rodents.txt, the project's staged copy 2cd8664560dc805315ded8c8c55b95ecb6f1ee92 not-redistributable neuroscience

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them the methods available for studying perceptual decisions in mice and rats, and whether they can reach the rigour established in primate psychophysics Quotation withheld because this record is not cleared for redistribution. can perceptual decision-making be probed in rodents to a standard comparable with primates, and what does each task design let one conclude Quotation withheld because this record is not cleared for redistribution. sets out what each rodent task design can and cannot rule out, and corrects two widely held beliefs about rodents that turn out to be artefacts of laboratory practice Quotation withheld because this record is not cleared for redistribution. method Quotation withheld because this record is not cleared for redistribution. perceptual decision-making in rodents Quotation withheld because this record is not cleared for redistribution. review review Quotation withheld because this record is not cleared for redistribution. mechanisms a shift in decision criterion or in willingness to act inflates the response rate, which is then misread as an improvement in the neural representation of the stimulus Quotation withheld because this record is not cleared for redistribution. mechanisms free water over the weekend leaves animals under-motivated at the start of the week, which drives the observed day-of-week variation in trial counts and proficiency Quotation withheld because this record is not cleared for redistribution. a belief about the animal is traced back to how the animal is kept: mice look nocturnal because laboratory food is unlimited Quotation withheld because this record is not cleared for redistribution. the correction is checked against morphology rather than behaviour alone - the eyes lack the tapetum a nocturnal animal would have Quotation withheld because this record is not cleared for redistribution. task designs are ordered by what each one can rule out, so the choice of design is presented as a choice of which confound survives Quotation withheld because this record is not cleared for redistribution. the criterion problem is made concrete with a hypothetical optogenetic result that would be misread under a go/no-go design Quotation withheld because this record is not cleared for redistribution. d-prime is preferred over the raw hit-minus-false-alarm difference specifically because the latter moves when only the criterion moves Quotation withheld because this record is not cleared for redistribution. a two-stimulus design removes the criterion entirely, because presence must be assigned to one location and absence to the other Quotation withheld because this record is not cleared for redistribution. one stimulus intensity is shown to be uninterpretable: sensitivity, bias and lapse rate produce identical data at a single point Quotation withheld because this record is not cleared for redistribution. the fix is a full psychometric curve, whose slope, position and tails separate the three Quotation withheld because this record is not cleared for redistribution. short training is argued against on the grounds that error trials become uninterpretable, since a mistake may be about the stimulus or about the task Quotation withheld because this record is not cleared for redistribution. a well-trained animal makes error trials interpretable, and near-perfect performance on easy trials is the check that the task itself is understood Quotation withheld because this record is not cleared for redistribution. an unpublished four-fold day-of-week effect on trial count is reported, traced to the weekend water schedule Quotation withheld because this record is not cleared for redistribution. the tacit knowledge attached to a standard task is named as knowledge that circulates informally and appears in no paper Quotation withheld because this record is not cleared for redistribution. The authors state that their own background biases the review toward one brain region and one modality. what this limits: the review's coverage: DUE TO OUR OWN BACKGROUNDS AND INTERESTS, WE PERHAPS UNDULY EMPHASIZE ONE BRAIN REGION, THE CEREBRAL CORTEX, AND ONE SENSORY MODALITY, VISION Quotation withheld because this record is not cleared for redistribution. The review is framed around unresolved questions rather than settled answers about rodent perceptual ability. what this limits: everything in the review: WE HIGHLIGHT WHAT ARE MAINLY OPEN QUESTIONS ABOUT RODENT PERCEPTUAL ABILITIES AND OUR ABILITY TO ASSESS THESE EXPERIMENTALLY Quotation withheld because this record is not cleared for redistribution. The widely held view that rats out-train mice is flagged by the authors as untested. what this limits: any reliance on the rats-versus-mice trainability view: AS FAR AS WE KNOW, HOWEVER, THIS VIEW HAS NOT BEEN DIRECTLY TESTED Quotation withheld because this record is not cleared for redistribution. the widely held view that rats out-learn mice has never been directly tested what this limits: the same scope as limitations[2] - the view is widely held and untested Quotation withheld because this record is not cleared for redistribution. and may be an artefact of training technique rather than a species difference what this limits: the species comparison: RATS MAY ONLY APPEAR TO BE EASIER TO TRAIN SIMPLY BECAUSE MICE REQUIRE DIFFERENT TRAINING TECHNIQUES Quotation withheld because this record is not cleared for redistribution. a term the field uses to justify design decisions has no agreed definition what this limits: every design decision justified by overtraining: THE NOTION OF OVERTRAINING, HOWEVER, IS POORLY DEFINED Quotation withheld because this record is not cleared for redistribution. whether longer training adds more uncontrolled plasticity than shorter training is not established either way what this limits: the argument for shorter training: IT IS NOT CLEAR THAT THIS PLASTICITY WOULD COMPLICATE INTERPRETATION of behavioral and neural data Quotation withheld because this record is not cleared for redistribution. virtual environments lack vestibular feedback, which visual feedback may not compensate for what this limits: virtual-environment results: THE VISUAL FEEDBACK IN THE VIRTUAL ENVIRONMENT MAY NOT ENTIRELY MAKE UP FOR THE LACK OF VESTIBULAR FEEDBACK Quotation withheld because this record is not cleared for redistribution. rodents should not be treated as a replacement for primates what this limits: the use of rodents as a substitute: IT WOULD BE A MISTAKE TO CONSIDER RODENTS AS A COMPLETE REPLACEMENT FOR PRIMATES IN INVESTIGATIONS OF PERCEPTUAL DECISION-MAKING Quotation withheld because this record is not cleared for redistribution. whether rodent designs can reach primate psychophysical rigour is left open what this limits: the future of rodent psychophysics: WILL EXPERIMENTAL DESIGNS IN RODENTS BE ABLE TO TAKE ADVANTAGE OF RIGOROUS primate methods - left open Quotation withheld because this record is not cleared for redistribution. punishment-based training learns fast but may induce stress that blocks harder tasks what this limits: punishment-based training: THEY MAY CREATE STRESS, WHICH MIGHT PREVENT LEARNING OF MORE COMPLEX TASKS Quotation withheld because this record is not cleared for redistribution. the whole task toolbox - random-dot motion, forced choice, psychometric curves - carried into rodents together with the standards for interpreting it Quotation withheld because this record is not cleared for redistribution. the separation of sensitivity from criterion, imported as the reason a rodent task must measure false alarms and not only hits Quotation withheld because this record is not cleared for redistribution. the mouse as a nocturnal, minimally visual animal versus as a facultatively diurnal one - the first turns out to be an artefact of unlimited laboratory food Quotation withheld because this record is not cleared for redistribution. detectability - as the raw difference between hits and false alarms versus as d-prime, which is criterion-independent; the two diverge whenever the criterion moves Quotation withheld because this record is not cleared for redistribution. a performance change - as reduced sensitivity, as shifted bias, or as raised lapse rate; three distinct mechanisms indistinguishable at a single stimulus intensity Quotation withheld because this record is not cleared for redistribution. two-alternative choice (one stimulus, two responses) versus two-alternative forced choice (two stimuli, two responses) - names that are often conflated and designs that differ in exactly which confound they remove Quotation withheld because this record is not cleared for redistribution. an error trial - as misperception of the stimulus versus as confusion about the task; only training separates them Quotation withheld because this record is not cleared for redistribution. a species difference in trainability versus a difference in the training techniques that happen to have been developed Quotation withheld because this record is not cleared for redistribution. rodent frontal cortex as fundamentally different from primate on histological grounds versus as functionally equivalent on behavioural grounds Quotation withheld because this record is not cleared for redistribution. claim_type: method a second reading reaching the same value. Recorded: method — note: a second reading reaching the same value from a different sentence. Not applied - a person decides. Quotation withheld because this record is not cleared for redistribution. method: narrative methods review of task designs, sensory environments, training regimes and reward schedules, evaluated against signal detection theory a second reading DIFFERING from the recorded value. Recorded: review — note: a second reading, verbatim-verified, disagreeing with the value on the record. Not applied - a person decides. Quotation withheld because this record is not cleared for redistribution. phenomenon: perceptual decision-making - using ambiguous sensory information to guide action a second reading DIFFERING from the recorded value. Recorded: perceptual decision-making in rodents — note: a second reading, verbatim-verified, disagreeing with the value on the record. Not applied - a person decides. Quotation withheld because this record is not cleared for redistribution. critiques → the view that rodent behaviour is only weakly influenced by vision Quotation withheld because this record is not cleared for redistribution. contradicts → the belief that mice are nocturnal, which the paper traces to unlimited laboratory food Quotation withheld because this record is not cleared for redistribution. critiques → rodent studies that measure behaviour at a single stimulus intensity differs in: how many stimulus intensities were measured: OBTAINING BEHAVIORAL DATA AT ONLY A SINGLE INTENSITY (AS IS OFTEN DONE IN RODENT STUDIES) WOULD NOT ALLOW ONE TO DISTINGUISH THE EFFECTS OF SENSITIVITY, BIAS, AND LAPSE RATE - one measurement cannot separate three quantities Quotation withheld because this record is not cleared for redistribution. critiques → the untested consensus that rats learn perceptual tasks faster than mice status: the paper's own words: not been directly tested Quotation withheld because this record is not cleared for redistribution. critiques → the use of the difference between hit and false-alarm rates as a detectability measure differs in: whether the measure survives a criterion shift: THE LATTER CAN CHANGE MARKEDLY FOLLOWING SIMPLE CHANGES IN CRITERION, where the recommended measure does not Quotation withheld because this record is not cleared for redistribution. builds-on → primate psychophysics, whose task toolbox and standards the paper takes as the benchmark Quotation withheld because this record is not cleared for redistribution. supports → evidence that rodent frontal cortex, like primate, supports working memory and categorization, against the histological argument for fundamental difference Quotation withheld because this record is not cleared for redistribution. Signal Detection Theory —contrasted-with→ Drift Diffusion Model Signal Detection Theory Drift Diffusion Model differs in: THE AXIS IS WHETHER THE DECISION VARIABLE EVOLVES WITH ACCUMULATING EVIDENCE. 'An increasingly popular ALTERNATIVE TO SDT, the Drift Diffusion Model Quotation withheld.' Both models produce a choice from noisy evidence; only one gives the decision a trajectory in time, which is what makes reaction times predictable and is the whole of the difference. · strength of the relation: 1 Quotation withheld because this record is not cleared for redistribution. go/no-go —contrasted-with→ two-alternative choice go/no-go two-alternative choice strength of the relation: 1 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but endpoint by anaphora: 'this design' = the two-alternative choice task (previous sentences). Quotation withheld because this record is not cleared for redistribution. two-alternative choice —contrasted-with→ two-alternative forced choice two-alternative choice two-alternative forced choice strength of the relation: 0.95 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but endpoint by anaphora: the refinement is of the two-choice design (previous section); the span names 'two-alternative forced choice' only. Quotation withheld because this record is not cleared for redistribution. criterion —distinct-from→ two-alternative forced choice criterion two-alternative forced choice strength of the relation: 0.95 Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span argues two-interval designs remove presence/absence bias; distinct-from 'criterion' is loose. Quotation withheld because this record is not cleared for redistribution. psychophysics —operationalised-as→ psychometric curves psychophysics psychometric curves strength of the relation: 0.95 Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span describes obtaining psychometric functions; operationalised-as is loose. Quotation withheld because this record is not cleared for redistribution. water control —causes→ motivation water control motivation strength of the relation: 0.9 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but endpoint by anaphora: 'this regime' = the regime of water control (previous sentence). Quotation withheld because this record is not cleared for redistribution. Posterior Parietal Cortex —causes→ evidence integration Posterior Parietal Cortex evidence integration strength of the relation: 0.7 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the PPC is 'POISED TO' integrate multisensory inputs; causes is hedged. Quotation withheld because this record is not cleared for redistribution. lapse rate —part-of→ psychometric curves lapse rate psychometric curves strength of the relation: 0.95 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span writes 'psychometric function', not 'psychometric curves'; part-of is loose. Quotation withheld because this record is not cleared for redistribution. Go/no-go tasks are highly vulnerable to changes in the animal's motivation and response criterion. Quotation withheld because this record is not cleared for redistribution. Requiring a decision on every trial makes the design immune to changes in the subject's willingness to respond. Quotation withheld because this record is not cleared for redistribution. With two stimulus presentations subjects cannot be biased toward presence or absence, since each choice assigns presence to one location and absence to the other. Quotation withheld because this record is not cleared for redistribution. Data from a single stimulus intensity, as often collected in rodent studies, cannot separate the effects of sensitivity, bias and lapse rate. Quotation withheld because this record is not cleared for redistribution. Mice are reliably nocturnal only when food is unlimited, as in the laboratory; with scarcer food they sleep part of the night to save energy. Quotation withheld because this record is not cleared for redistribution. Rodent eye anatomy provides independent evidence against the nocturnality assumption. Quotation withheld because this record is not cleared for redistribution. Performance can differ fourfold across the week - about 100 trials per session on Mondays against about 400 on Fridays - by the authors' unpublished observations. Quotation withheld because this record is not cleared for redistribution. Despite the technical advantages of rodents, the review argues against treating them as a substitute for primate work on perceptual decisions. Quotation withheld because this record is not cleared for redistribution. mice are reliably nocturnal only when food is unlimited, as it is in the laboratory Quotation withheld because this record is not cleared for redistribution. laboratory mice can become entirely diurnal in nature, and rodent eyes lack the tapetum expected of a nocturnal animal Quotation withheld because this record is not cleared for redistribution. mouse spatial resolution is about 100 times lower than primate, yet vision is the primary sense rodents use to locate themselves Quotation withheld because this record is not cleared for redistribution. mice performed about 100 trials per session on Mondays against about 400 on Fridays Quotation withheld because this record is not cleared for redistribution. contemporary regimes fix this with constant daily intake and seven-day training Quotation withheld because this record is not cleared for redistribution. some rodent tasks take 2 to 10 days to teach and others 3 to 6 weeks, and rodent training rarely exceeds 2 months Quotation withheld because this record is not cleared for redistribution. rats combine multisensory information about as optimally as humans do Quotation withheld because this record is not cleared for redistribution. even within mice, body weight correlates strongly with the number of trials performed per day Quotation withheld because this record is not cleared for redistribution. go/no-go task go/no-go task Quotation withheld because this record is not cleared for redistribution. two-alternative forced choice two-alternative forced choice Quotation withheld because this record is not cleared for redistribution. detectability detectability Quotation withheld because this record is not cleared for redistribution. Signal Detection Theory Signal Detection Theory Quotation withheld because this record is not cleared for redistribution. Drift Diffusion Model Drift Diffusion Model Quotation withheld because this record is not cleared for redistribution. go/no-go go/no-go Quotation withheld because this record is not cleared for redistribution. two-alternative choice two-alternative choice Quotation withheld because this record is not cleared for redistribution. forced choice forced choice Quotation withheld because this record is not cleared for redistribution. Drift Diffusion Model Drift Diffusion Model Quotation withheld because this record is not cleared for redistribution. psychometric function psychometric function Quotation withheld because this record is not cleared for redistribution. d d Quotation withheld because this record is not cleared for redistribution. perceptual sensitivity perceptual sensitivity Quotation withheld because this record is not cleared for redistribution. sensory bias sensory bias Quotation withheld because this record is not cleared for redistribution. lapse rate lapse rate Quotation withheld because this record is not cleared for redistribution. ~100 times — how much lower spatial resolution is in rats and mice than in primates Quotation withheld because this record is not cleared for redistribution. at least 10 — number of retinotopic areas in the mouse visual cortex Quotation withheld because this record is not cleared for redistribution. 2–10 days / 3–6 weeks — range of training times needed for mice on easier versus harder perceptual tasks Quotation withheld because this record is not cleared for redistribution. 6–12 months — typical training time for a primate to reach proficiency on complex tasks, for contrast with rodents Quotation withheld because this record is not cleared for redistribution. a factor of 4 (~100 vs. ~400 trials/session) — Monday-versus-Friday difference in trials completed under a 5-day water-control regime Quotation withheld because this record is not cleared for redistribution. 25 ml/kg/day — target constant daily water intake in contemporary training regimes Quotation withheld because this record is not cleared for redistribution. 10 million years — how much more recently rodents and primates share a common ancestor than either does with the cat Quotation withheld because this record is not cleared for redistribution. ~100 times lower — rodent spatial resolution against primate Quotation withheld because this record is not cleared for redistribution. factor of 4 — Monday against Friday trial counts under the weekend water regime Quotation withheld because this record is not cleared for redistribution. 2-10 days or 3-6 weeks — rodent training durations, rarely beyond 2 months Quotation withheld because this record is not cleared for redistribution. 6-12 months — primate training to proficiency on complex tasks Quotation withheld because this record is not cleared for redistribution. 25 ml/kg/day — a contemporary constant water-intake target Quotation withheld because this record is not cleared for redistribution. Signal Detection Theory Signal Detection Theory Quotation withheld because this record is not cleared for redistribution. false alarms false alarms Quotation withheld because this record is not cleared for redistribution. Drift Diffusion Model Drift Diffusion Model Quotation withheld because this record is not cleared for redistribution. lapse rate lapse rate Quotation withheld because this record is not cleared for redistribution. stimulus strength stimulus strength Quotation withheld because this record is not cleared for redistribution. decision criterion decision criterion Quotation withheld because this record is not cleared for redistribution. trials/session trials/session Quotation withheld because this record is not cleared for redistribution. training duration training duration Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1038/nn.3410.