
Fish detour tests: a side preference may not persist
Ten turns in a maze can produce a strong side preference without establishing a lasting individual trait. A 2020 study found that detour-test preferences changed substantially between sessions. Its results make repeatability a prerequisite for interpreting left–right scores as stable behavioural lateralisation.
- Content type
- Scientific news
- Sector
- Research facilities
- Animal group
- Fish
- Keywords
- EthogramStressTemperature
Ten choices describe a session, not a lifetime
In a directional detour test, a fish approaches a barrier and turns left or right. The balance of choices is used to calculate a lateralisation index, often with the intention of inferring behavioural asymmetry or preferential eye use. That interpretation depends partly on whether the same individual shows a similar preference when tested again.
Roche and colleagues addressed this question in a 2020 study combining new experiments in four fish species with a reanalysis of published guppy data. The experiments examine whether a particular measurement supports the interpretation assigned to it.
The question is not whether fish can have brain asymmetries or eye-use preferences. It is whether this apparatus and its scores reliably identify persistent differences between individuals. A choice observed in one session and a stable individual phenotype are different kinds of information for an experiment.
Repeat the measurement in identified animals
The new experiments involved goldsinny wrasse, Ctenolabrus rupestris; the reef fishes Neopomacentrus azysron and Pomacentrus amboinensis; and zebrafish, Danio rerio. The analysed samples were 57, 52, 60 and 49 individuals respectively, totalling 218. These are the analysed sample sizes, which should be distinguished from the numbers initially collected.
Each fish was scheduled for four series of ten choices at forty-eight-hour intervals. Some series remained incomplete because individuals did not continue through all passages. Ten movements by the same fish within a session are not ten independent biological replicates.
The researchers also reanalysed forty female guppies, Poecilia reticulata, previously tested with neutral or social stimuli. Those data extend the species comparison, but they are not a fifth newly collected sample tested on precisely the same schedule as the other four species.
For anyone assessing an experimental report, this distinction matters. The number of choices, completed sessions and individuals should all be recoverable. Pooling those counts into a large total can conceal the level at which the biological question is being tested.
Direction and strength are different summaries
A relative index retains the direction of a preference. Predominantly right and predominantly left choices produce opposite signs. An absolute index removes the sign and describes the strength of the imbalance. These summaries consequently answer different questions when averaged across a group.
A relative group mean near zero can combine strong but opposing individual preferences. Conversely, a high absolute score in one session does not establish that the individual will retain its direction in another session. A report needs to identify the index and the level being interpreted: choice, session, individual or population.
Repeatability describes how much variation is associated with persistent differences between individuals under the conditions studied. For relative lateralisation in the four new species, the estimates were very low, ranging from 0.006 to 0.028. The guppy reanalysis yielded estimates of 0.045 or 0.076 according to stimulus. These values are not success rates or probabilities that a fish has a preferred side.
A strong session preference could change later
Some individuals produced strongly directional scores in a series and different preferences on another day. This explains why sorting animals into enduring categories using one session can be misleading. Selecting the most apparently right-biased fish today does not ensure that the same animals will retain that ranking on a subsequent test.
The paper also reviews statistical practices used in earlier studies and proposes approaches suited to binary choices. A statistically significant difference between sets of scores does not, by itself, demonstrate the individual stability that an investigator wants to interpret. The biological claim needs to be defined before choosing the numerical summary.
For a new project, retaining each choice, fish identity, session and incomplete series makes the measurement easier to assess. Exclusions and stopping criteria should be specified before analysis. Repeated testing needs a proportionate scientific purpose; it should not be extended simply until a conspicuous preference emerges.
Check stability before interpreting an exposure effect
Before attributing a lateralisation change to stress or an exposure, a team needs evidence about test stability in reference conditions. Temperature, lighting, barrier geometry and observer intervention should be documented. Recording them describes the measurement context; it does not prove that any particular factor caused the findings.
The conclusions remain tied to the tested species, stages and assay variants. A detour around a transparent obstacle to reach a reward is not interchangeable with a left–right choice at a barrier. Nor are mirror and circular-arena tests automatically validated by criticism of this apparatus.
The practical decision is to require evidence of repeatability before treating a session score as a persistent individual characteristic. Where that evidence is missing, a report can describe the observed preference, its conditions and its uncertainty without assigning the fish a lasting temperament or laterality category. The study does not show that all fish behaviour is unstable.
Vetofish can help teams define observational units and critically assess behavioural indicators through its training programmes, including support for aquatic research facilities.


