This repository contains raw data and analysis code for the manuscript entitled "Once an Optimist, Always an Optimist? Studying Cognitive Judgment Bias in Mice" from Marko Bračić, Lena Bohn, Viktoria Siewert, Vanessa von Kortzfleisch, Holger Schielzeth, Sylvia Kaiser, Norbert Sachser, S. Helene Richter, accepted for publication in the journal Behavioral Ecology.

The aim of the study was to investigate the causes and stability of cognitive judgment bias (aka "optimism").

Individuals differ in the way they judge ambiguous information: some individuals interpret ambiguous information in a more optimistic, and others in a more pessimistic way. Over the past two decades, such “optimistic” and “pessimistic” cognitive judgement biases (CJBs) have been utilized in animal welfare science as indicators of animals’ emotional states. However, empirical studies on their ecological and evolutionary relevance are still lacking.

We, therefore, aimed at transferring the concept of “optimism” and “pessimism” to behavioral ecology and investigated the role of genetic and environmental factors in modulating CJB in mice, using an automated, touchscreen-based active choice paradigm. In addition, we assessed the temporal stability of individual differences in CJB.

We show that the chosen genotypes (C57BL/6J and B6D2F1N) and environments (“scarce” and “complex”) did not have a statistically significant influence on the responses in the CJB test. By contrast, they influenced anxiety-like behavior (assessed in the elevated plus maze (EPM), an open field test (OFT), and a free exploration test (FET)) with C57BL/6J mice and mice from the “complex” environment displaying less anxiety-like behavior than B6D2F1N mice and mice from the “scarce” environment. As the selected genotypes and environments did not explain the existing differences in CJB, future studies might investigate the impact of other genotypes and environmental conditions on CJB, and additionally, elucidate the role of other potential causes like endocrine profiles and epigenetic modifications. Furthermore, we show that individual differences in CJB were repeatable over a period of seven weeks, suggesting that CJB represents a temporally stable trait in laboratory mice. Therefore, we encourage the further study of CJB within an animal personality framework.

The experiment lasted for one year (February 2019 – March 2020) and was conducted in two independent batches. Laboratory mice were obtained from a professional breeder at the age of four weeks and housed at the University of Münster. We housed female mice of two strains/genotypes (C57BL/6J and B6D2F1N) in two environmental conditions: a “scarce environment” and a “complex environment”. Mice were housed in same-strain groups of three individuals per cage. To allow for individual identification within cages, all mice received partial ear punches (earcut) upon arrival. The three mice housed in the same cage were exposed to the same environmental conditions but participated in different sets of the following experimental phases: a touchscreen training phase (TS), first cognitive judgement bias (CJB) test phase, repeated CJB testing phase, and behavioral test phase.

One mouse from each cage was touchscreen trained and tested in the first CJB test phase before it was removed from the cage to participate in another experiment. The second mouse was touchscreen trained as well but was then repeatedly tested in the CJB test (four CJB tests in total) and afterward entered the behavioral test phase. The third mouse in a cage participated only in the behavioral test phase.

With the data from the three experimental phases, we created three different datasets contained in this repository, each with a different sample size: 

• CJB_genotype-environment.txt contains source data used to analyse the influences of genotype and environment on CJB (first cognitive judgement bias (CJB) test phase, N = 39)
• CJB_repeatablity.txt contains source data used to analyse the repeatability of CJB (repeated CJB testing phase, N = 19)
• behavioural_tests.txt contains source data used to analyse the influences of genotype and environment on anxiety-like behaviour and spatial learning (behavioral test phase, N = 36)

Cognitive judgment bias (CJB) test

The choice score obtained from the CJB test served as a relative behavioral indicator of CJB. The choice score was calculated based on the mice’s responses (optimistic or pessimistic) during 5 testing sessions (days) for each individual for each cue (positive, near positive, middle, and near negative, negative).

Formula: Choice Score = (N choices ("optimistic") - N choices ("pessimistic")) / ( N choices ("optimistic" + "pessimistic"))

Data were automatically recorded by the ABET II software (version 2.20., Campden Instruments Ltd., Loughborough, Leics., UK) and the choice score was calculated in R.

As individuals needed different numbers of touchscreen training sessions before the CJB test, they were tested on different dates. The location of the correct side for the cues was counterbalanced between mice (trial_type): e.g., one mouse per cage had to touch the right-hand side in response to the positive cue to get a big reward, while the other mouse had to touch the left-hand side in response to the positive cue. 

Behavioral tests

The test battery included an elevated plus maze (EPM), an open field test (OFT), and a free exploration test (FET) to assess anxiety-like behavior and exploration. The subsequent labyrinth maze (LM) served to measure spatial learning. Tests were video recorded (Logitech Webcam Pro 9000) and automatically tracked (ANY-maze, version 5.33, Stoelting Co., Wood Dale, IL, USA).

Data analysis

Data were analysed by using linear mixed-effects models (LMM) in R (version 4.1.2) . Data analysis can be replicated with provided R scripts.

Variables in the dataset are explained in the ReadMe file and further clarified in the methods above (variables marked bold). Due to a setup error in the labyrinth maze (LM), three mice had to be excluded from the (labyrinth maze) analysis (NA).

For the analysis of each dataset, a separate R script is provided. To successfully run the R script, make sure that the data files are in R working directory (e.g. keep data and script in the same folder or change the working directory to the folder where data is saved) and that listed R packages are installed.