Table 1.

Predictor variables and model outputs for GLMMs (pecks: models 1, 2, 3b and reward worm latencies: model 3a, c), and GLM (reward worm latencies: model 4; pecks: model 5)

Models	Free worm	Sex	Body condition	Treatment	Trial
(1) Cylinder 1: opaque vs transparent	0.018 ± 0.038; χ2 = 0.222, p = 0.638	0.030 ± 0.262; χ2 = 0.013, p = 0.301	0.206 ± 5.780; χ2 = 0.305, p = 0.581	n/a	− 0.231 ± 1.366; χ2 = 172.798, p < 0.001
(2) Cylinder 1: transparent [improvement across trials]	0.022 ± 0.028; χ2 = 0.605, p = 0.437	0.304 ± 0.163; χ2 = 3.43, p = 0.064	2.376 ± 3.638; χ2 = 0.424, p = 0.515	n/a	− 0.560 ± 0.156; χ2 = 12.167, p < 0.001
(3a) No-barrier habituation	n/a	− 2.638 ± 4.531; χ2 = 0.355, p = 0.551	− 53.045 ± 101.676; χ2 = 0.285, p = 0.593	− 0.961 ± 4.491; χ2 = 0.049, p = 0.825	− 11.993 ± 1.459; χ2 = 57.935, p < 0.001
(3b) Response training: pecks	n/a	− 0.144 ± 0.207; X2 = 0.48, p = 0.487	− 8.209 ± 4.678; χ2 = 2.98, p = 0.084	1.146 ± 0.200; χ2 = 363.73, p < 0.001	− 0.600 ± 0.029; χ2 = 363.73, p < 0.001
(3c) Response training: reward worm	n/a	− 9.426 ± 4.890; χ2 = 3.85, p = 0.050	− 21.760 ± 109.206; χ2 = 0.042, p = 0.837	12.746 ± 4.784; χ2 = 7.159, p = 0.007	− 8.183 ± 0.588; χ2 = 166.764, p < 0.001
(4) Shortcut [treatment = through vs around barrier]	n/a	− 2.660 ± 5.098; χ2 = − 0.522, p = 0.604	− 48.186 ± 116.806; χ2 = − 0.413, p = 0.681	7.094 ± 6.140; χ2 = 1.155, p = 0.253	n/a
(5) Cylinder 2: retest [post training improvement]	0.067 ± 0.569; χ2 = 0.015, p = 0.902	− 4.368 ± 4.647; χ2 = 0.954, p = 0.329	− 24.668 ± 103.093; χ2 = 0.062, p = 0.803	18.496 ± 4.533; χ2 = 15.886, p < 0.001	n/a

Estimates ± SEM are presented with their corresponding Chi squared (χ²) and significance values (p). n/a = variable not included in analysis