Results

Experiment 1: Effect of SB242084 on Egocentric Reversal Learning
Four of the 72 animals were excluded from analysis. Three of these animals failed to respond in the spatial discrimination test, and the remaining one was excluded after becoming ill. An additional two animals failed to complete the full reversal test within 250 trials (one in each drug condition). Animals assigned to the two maze configurations, three test conditions, or two drug groups did not differ in performance in the spatial discrimination (Table 1). There were no significant main effects of drug, test condition, or drug × test condition interaction on probe-trials to criterion (M = 1.5±0.07).
10.1371/journal.pone.0077762.t001 Table 1  Mean trials and incorrect responses (± SEM) to criterion in spatial discrimination in the SB242084 and 5-HT2CR experiments.     SB242084 improved performance in the perseverance test, but retarded performance in the learned non-reward test (Fig. 2A, B). SB242084 failed to affect learning in the full reversal test. On trials to criterion (Fig. 2A), there was a significant main effect of test condition (F2,61 = 13.10, p<.0001). Animals required more trials to reach criterion in the full reversal test than in the perseverance (p<.001) and learned non-reward tests (p<.0001). There was also a near significant drug × test condition interaction (F2, 61 = 3.13, p = .051). Separate one-way ANOVAs showed that SB242084 decreased trials to criterion in the perseverance test (F1,20 = 4.54, p<.05), while increasing trials to criterion in the learned non-reward test (F1,22 = 4.44, p<.05). SB242084 had no effect on trials to criterion in the full reversal test (p = .20).
10.1371/journal.pone.0077762.g002 Figure 2  Performance in the full reversal, perseverance and learned non-reward tests in the SB242084 (a–b) and 5-HT2CR KO experiments (c–d).
Asterisk denote differences at which p<.05. (a) Significant main effect of test (F2,61 = 13.10, p<.0001) and a near significant drug × test interactions (F2, 61 = 3.13, p = .051) on trials to criterion. SB242084 decreased trials to criterion in the perseverance test (F1,20 = 4.54, p = .046) and increased trials to criterion in the learned non-reward test (F1,22 = 4.44, p = .047). (b) Significant main effects of test condition (F2, 61 = 9.65, p<.0001) and drug × test interactions (F2, 61 = 3.46, p = .037) on incorrect responses to criterion. SB242084 decreased incorrect responses to criterion in the perseverance test (F1,20 = 5.96, p = .024). (c-d) Significant main effects of genotype on trials (F1,26 = 5.83, p = .023) and incorrect responses (F1,26 = 4.45, p<.045) to criterion in the perseverance test. Genotype had no significant effects on performance in the full reversal and learned non-reward tests.   On incorrect trials to criterion (Fig. 2B), there was a significant main effect of test condition (F2, 61 = 9.65, p<.0001) and drug × test condition interaction (F2, 61 = 3.46, p<.05). Animals made more incorrect responses to criterion in the full reversal test than in the perseverance (p<.01) and learned non-reward tests (p<.0001). SB242084 decreased the number of incorrect responses made in the perseverance test (F1, 20 = 5.96, p<.05). Although the difference failed to reach significance, SB242084 treated animals made more incorrect responses in the learned non-reward test (F1, 22 = 2.97, p = ns). There were no significant effects of drug or drug × test condition interaction on early and late errors to criterion (p>.05; Table 2).
10.1371/journal.pone.0077762.t002 Table 2  Early and late errors to criterion in the three test-conditions of Experiment 1 and 2.      Experiment 1. No significant effects of drug or drug × test condition interaction. Significant main effect of test condition on early (F2,61 = 5.37, p<.01) and late errors (F2,61 = 10.22, p<.0001). Animals made more early and late errors in the full reversal condition than the perseverance and learned non-reward conditions (p<.05). Experiment 2. No significant effects of genotype.

Experiment 2: Effect of 5-HT2cR KO on Egocentric Reversal Learning
There were no effects of genotype on probe-trials to criterion (Grand mean = 1.4±0.06). No animals failed to complete the initial spatial discrimination or the learned non-reward test. However significantly more 5-HT2CR KO animals (N = 8) than WT animals (N = 2) failed to complete either the full reversal or perseverance tests (Table 3; χ2 = 7.53, p<.01).
10.1371/journal.pone.0077762.t003 Table 3  Proportion (%) of WT and 5-HT2CR KO mice reaching criterion in egocentric discrimination, full reversal, perseverance and learned non-reward tests.     Significantly fewer 5-HT2CR KO animals than WT animals reached criterion in the perseverance condition (x2 = 4.2, p = .04).   5-HT2CR KO animals required more trials to criterion (F1,31 = 6.08, p<.05) and made more incorrect responses to criterion (F1,31 = 6.11, p<.05) in egocentric discrimination learning (Table 1). There was also a non-significant trend for 5-HT2CR KO animals to perform worse in the full reversal test by requiring more trials (F1,31 = 3.96, p = ns) and making more incorrect responses to criterion (Fig. 2C, D; F1,31 = 3.74, p = ns).
In the perseverance test, there was a significant effect of genotype on trials (F1,26 = 5.83, p<.05) and incorrect responses to criterion (F1,26 = 4.45, p<.05) with 5-HT2CR KO mice showing retarded learning relative to WTs. Although 5-HT2CR KO mice tended to perform better than WTs in the learned non-reward test, these differences were non-significant. There were no significant effects of genotype on early and late errors to criterion (p>.05; Table 2).
To further explore if the performance in the full reversal and perseverance tests could be accounted for by differences in initial discrimination learning, the data for these two test conditions were re-analysed using the initial spatial discrimination performance as a covariate. The near-significant effects of genotype on trials (Fig. 2C) and incorrect responses (Fig. 2D) to criterion in the full reversal test could be accounted for by retarded egocentric discrimination learning (trials, F1,30 = 0.64, p = .43; incorrect responses, F1,30 = 0.57, p = .46). However, the effect of genotype on performance in the perseverance test (Fig. 2C, D) remained statistically significant (trials, F1,25 = 7.65, p = .01; incorrect responses, F1,25 = 4.23, p = .05).

Experiment 3: Effect of SB242084 on Unrewarded Choice Behaviour
Animals spent more time in the novel arm and made more arm-entries into the novel arm, and SB242084 had no effect on either of these measures (Table 4). There were no effects of maze-configuration or drug × maze-configuration interaction on entries into the novel or old arms. There were significant effects of phase (pre- and post-change) on proportion of time (F1,26 = 12.27, p<.01) and proportion of arm-entries made into the novel arm (F1,26 = 34.78, p<.0001). There were no effects of drug (p ≥.16) or drug × phase interactions (p ≥.19) on proportion of time spent in the novel arm or proportion of arm entries made into the novel arm.
10.1371/journal.pone.0077762.t004 Table 4  Proportion of time and proportion of entries (± SEM) before a 45° shift (pre-shift) and after a 45° shift (post-shift) in vehicle and SB242084 treated animals.     **  p<.01.

Di