Table 2.

Optimum classification results in the DNN analysis

Test Phase	Accuracy	Description
Phase 1 (NWT = 13, NAppNL-G-F/NL-G-F = 14)	89.3% ± 9.8%	1[Delay-discounting] Lick rate at saccharin bottles during 6-s delay∗ 2[PP learning] Visit rate to the correct corner chamber on day 1∗ 3[Delay-discounting] Nose-poke rate at doorways to access the saccharin solution during 1-s delay 4[Delay-discounting] Nose-poke rate at doorways to access the saccharin solution during 5-s delay∗
Phase 2 (NWT = 14, NAppNL-G-F/NL-G-F = 15)	100% ± 0.0%	1[PPR learning] Visit rate to the correct corner chamber on day 1∗ 2[Impulsivity] Premature trial rate on day 8 3[Attention] Correct rate from total non-premature trial at stimulus duration of 0.5 s∗ 4[PPR learning] Visit rate to the correct corner chamber on day 2∗ 5[PA learning] Nose-poke error rate during extinction 6[PP learning] Visit rate to the correct corner chamber on day 4 7[PPR learning] Visit rate to the correct corner chamber on day 5
Phases 1 and 2 (NWT = 10, NAppNL-G-F/NL-G-F = 10)	100% ± 0.0%	1[PPR learning of phase 2] Visit rate to the correct corner chamber on day 2∗ 2[Attention of phase 2] Omission error rate at stimulus duration of 0.5 s 3[PPR learning of phase 2] Visit rate to the correct corner chamber on day 3∗ 4[PPR learning of phase 1] Visit rate to the correct corner chamber on day 3 5[PP learning of phase 2] Visit rate to the correct corner chamber on day 4

This table shows optimum classification results (i.e., accuracy) in the DNN analysis and selected features for genotype classification using behavioral data of test tasks during phases 1, 2, and both phases. Asterisks (∗) denote features showing significantly different averages between WT and App^{NL-G-F/NL-G-F} mice (two-sample t test; p < 0.05).

PP, place preference; PPR, place preference reversal; PA, place avoidance.