Deciding which offspring to feed is one of the most critical decisions parents make, for both parental and offspring fitness. Despite knowing much about what choices parents make, we know little about how parents choose. What we do know about how the brain integrates sensory evidence when choosing between options comes from laboratory studies and models. However, such studies may not adequately reflect decisions made in nature—with real-world complexity and consequences. Our naturalistic experiment on decision-making in 62 wild Parus major parents overcomes this issue. Decision speed was impacted by whether parents chose to feed a typically-preferred chick or not, offspring starvation risk, decision complexity, and parental sex. Parents regularly moved food between chicks before committing, suggesting that parents were not confident or made a mistake. Such decision changes were predicted by similar factors as speed. After moving food, parents were more likely to continue gathering evidence post-decision, and their next decision was slower. These results demonstrate several factors impacting cognition, and perhaps metacognition, in wild birds. More broadly, our study demonstrates how crucial evolutionarily relevant experiments in natural settings are.

GENERAL INFORMATION

This README.txt file was updated on 3 September 2025.

Paper associated with this archive:
Caro, S.M., Villareal, R.M., Velasco, A.C., van Mastrigt, T., van Oers, K., Hinde, C.A., & Hofmann, H., Decision-making in the wild: Urgency and complexity drive feeding decision speed and the likelihood of revising a choice in a sex-dependent manner in great tit (Parus major) parents.

DOI: XXXXXXXXX (will be updated upon manuscript acceptance and publication)\

Corresponding author: Shana Caro: shana_caro@post.harvard.edu

Author contributions to data collection and code:

SMC, ACV, TvM and CAH collected field data. ACV and RMV scored video data. SMC wrote the code.

ACCESS INFORMATION

Licenses/restrictions placed on the data or code: CC0 1.0 Universal (CC0 1.0) Public Domain Dedication
Data derived from other sources: NA
Recommended citation for this data/code archive: xxxxxxx

DATA & CODE FILE OVERVIEW

This data repository consist of two csv data files, twelve video files, and one R code file, and this README document, with the following data filenames and variables:

Data files

1. "Data_Archive_for_Decision_Making_Manuscript.csv"

Description: This data file contains information on feeding visits made by great tit (Paris major) parents. Each row is a unique feeding visit where the parent brought a prey item back to the nest and then fed a chick.

Variables

Nest_ID: which nest it is
Blinded.Film.File: the file name wherein the video of this feeding visit can be found
confidence.in.decision: whether or not parent changed their mind before feeding a chick - did parents place food in the mouth of chick A and then move it to chick B? This has two levels (did not move food, or moved food)
parental.decision.time: the time in seconds from when parents enter the nest to making a decision (placing food in the mouth of the chick that ultimately eats that food)
chicks.probed.after: Numerical measure of the number of different chicks that a parent dipped its beak into the mouth of, after feeding a chick on that same nest visit (possible range 0 to 7)
parent.sex: male or female.
prey.size: a categorical variable indicating whether prey items are big, medium, or small compared to the size of the adult's beak
number.of.begging.chicks.total: Numerical count of how many chicks begged on the feeding visit (range 0-7)
fed.chick.beg.rank: the postural begging rank of the fed chick compared to its nest mates on that feeding visit, with 1 being highest and 7 lowest. A score of 1 indicates that the chick was begging as much or more than its nestmates. Postural begging was coded from 0-3, with 3 being the highest intensity.
fed.chick.proximity.rank: the proximity rank of the fed chick compared to its nest mates on that feeding visit, with 1 being closest to the parent and 7 being farthest from the parent (possible range 1 to 7). In analyses, ranks 5, 6, and 7 are rare enough that they are collapsed into the value 5 (thus, the range in analyses is 1-5).
fed.chick.weight.rank: the weight rank of the fed chick compared to its nest mates. Chicks were weighed prior to filming, and ranked by body size (in grams, range 1-7, with 1 being the heaviest chick and 7 being the lightest chick).
chicks.probed.before.numerical.confidence.measure: the number of different chicks that the parent placed food into the mouths of, before moving it to another chick (possible range 0 to 6). Chicks do not eat when probed. A value of zero indicates that the parent only placed food in the mouth of the chick that was ultimately fed. A value of two would indicate the parent placed food in the mouth of Chick A, moved it to Chick B, and finally moved it to Chick C, who ate the food.
13.supplementation.treatment: a categorical variable indicating whether parents received extra food (yes) or not (no)
14.chicks.probed.before.feeding.in.previous.choice: on the parent's previous visit, the number of different chicks that the parent placed food into the mouths of, before moving it to another chick (possible range 0 to 6).
confidence.in.previous.choice: a categorical variable indicating whether, on the parent's previous feeding visit, any chick was probed (not confident) or not (confident)
unique.row.ID: a unique identifier for each feeding visit
Chick_ID: the identity of the chick that was fed on this feeding visit
Parent_ID: the identity of the parent on this feeding visit
visit.number.per.parent: the number of the parent's feeding visits scored. A value of 1 indicates it is the first visit scored for that parent, 2 indicates it is the second, etc.
nest.entrance.hour: the time (based on the timestamp in the video) that the parent entered the nest box
nest.entrance.minute: the time (based on the timestamp in the video) that the parent entered the nest box
nest.entrance.second: the time (based on the timestamp in the video) that the parent entered the nest box
decision.hour: the time (based on the timestamp in the video) that the parent placed food in the mouth of the chick that was fed
decision.minute: the time (based on the timestamp in the video) that the parent placed food in the mouth of the chick that was fed
decision.second: the time (based on the timestamp in the video) that the parent placed food in the mouth of the chick that was fed
multiprey_yn: a categorical variable indicating whether the parent brought back a single prey item or multiple prey items. Visits with multiple prey items were excluded from this study
exclude.due.to.missing.data.or.obscured.view: a categorical variable indicating whether this feeding visit should be excluded (yes) or not (no) from analyses due to missing data from the camera being obscured during the visit.

*Note: Missing values in the data are labeled NA.

2. "speediness.fitness.effects.csv"
Description: This data file contains information on chick mortality outcomes by nest and parents' mean decision speeds during feeding visits.
Variables:

Supplemented: a categorical variable indicating whether parents received extra food (Supplemented) or not (Unsupplemented)
Nest: which nest it is
parent.sex: whether the row refers to the male or female of this nest
concatenated_ID: the identity of the parent
number.dead.by.day.7: the number of chicks that died in this nest between hatching (day 0) and day 7 (range 0 to 5 chicks).
Individual.Mean.Speediness: the mean decision time in seconds for that individual bird, based on all its recorded feeding visits. Decision time is measured from when the adult enters the nest and can first observe the chicks to when it places food in the mouth of the chick that is ultimately fed.
any.mortality.by.day.7: a categorical variable indicating whether any chicks died (some dead chicks) or not (No mortality) in the first week after hatching
any.mortality.by.day.7.binomial: a binomial variable indicating whether any chicks died (1) or not (0) in the first week after hatching
number.dead.after.day.8: the number of chicks that died in this nest between filming day (day 8) and fledging (day 21) (range 0 to 7 chicks).
any.mortality.by.day.8: a categorical variable indicating whether any chicks died (some dead chicks) or not (No mortality) in the time after filming before fledging
any.mortality.by.day.7.binomial: a binomial variable indicating whether any chicks died (1) or not (0) in the time after filming before fledging
change.in.brood.size: the difference in between a nest's original brood size (the number of chicks that hatched in that nest) and foster brood size (the number of chicks that were fostered into that nest for filming). Chicks remained in their fostered broods through fledging. A negative value indicates that the fostered brood was smaller than the original brood. A positive value indicates that the fostered brood was larger than the original brood (range -3 to +3 chicks).

3. Video Files
Twelve representative video files have been uploaded to the data depository. The file names follow the format letter_number, where the letter refers to the blinded nest and the number refers to the order of the video files.

a_2.AVI
aa_3.AVI
b_2.AVI
bb_2.AVI
c_2.AVI
cc_4.AVI
d_2.AVI
dd_2.AVI
e_4.AVI
ee_2.AVI
f_4.AVI
ff_2.AVI

4. "Code_Archive_for_Decision_Making_Manuscript.R"

Description: R code for all statistical analyses on decision speed, moving food, probing, and mortality inthe manuscript. The name and reason for each analysis is indicated with each model. The order of the code is 1) Load packages; 2) Import, clean, and transform data (e.g., rename factor levels or center and scale variables); 3) GLMER models for speed, moving food and probing; 4) Path (PiecewiseSEM) Models; and 5) Fitness consequences. Code is annotated within the R code file. For questions about code, please contact shana_caro@post.harvard.edu.

SOFTWARE VERSIONS
Analysis is done in R, version 4.4.0 (2024-04-24) --
"Puppy Cup"

Packages with versions used in published analyses: lme4 1.1-35.3, lmerTest 3.1-3, ggplot2 3.5.1, piecewiseSEM 2.3.0, semPlot 1.1.6, lavaanPlot 0.8.1, lavaan 0.6-17, ape_5.8, performance_0.12.2, effects_4.2-2, carData_3.0-5, lubridate_1.9.3, forcats_1.0.0, stringr_1.5.1, purrr_1.0.2, readr_2.1.5, tidyr_1.3.1, tibble_3.2.1, tidyverse_2.0.0, readxl_1.4.3, corrplot_0.92, dplyr_1.1.4, caper_1.0.3, mvtnorm_1.2-4,MASS_7.3-60.2, Matrix_1.7-0

Funding and Animal Ethics statements

Funding: This research was supported by a WIAS fellowship awarded to SMC at Wageningen University and SEX SEL-VAR project number 334544 funded by a Marie Curie Career Integration Grant from the European Commission (FP7) awarded to CAH. This work was also funded by a University of Texas at Austin Stengl-Wyer Postdoctoral Fellowship for SMC and a Stengl-Wyer InSTInCT REU Fellowship to RMV. We thank NIOO-KNAW for additional funding.

Ethics: Permission to do fieldwork in the Boslust forest was granted by Stichting Geldersch Landschap en Kasteelen. Permission for animal experiments was granted by Central Authority for Scientific Procedures on Animals (CCD) of the Netherlands under project number AVD-801002017831 to KVO.

Data and code from: Decision-making in the wild: Urgency and complexity drive feeding decision speed and the likelihood of revising a choice in a sex-dependent manner in great tit (Parus major) parents

Data files

Abstract