Fig 1.
For comparison, DI for a classic version of the object displacement task with 5-min sample and 24 hr test (same object both trials). n = 8, *P < 0.05 to chance. Data in S5 Data. DI, discrimination index.
Fig 2.
A. Panel: Trial structures for the 3 different conditions (object identity changed from trial to trial). In the overlapping condition, 1 location remains constant across all sample trials and the test trial, the second location varies. The locations in the last sample trial and in the test trial are equal, thus only cumulative memory across trials will lead to a preference for the less often shown location. In the stable condition, the locations remain the same in all sample trials and 1 object is displaced in the test trial. In the random condition, the locations were pseudo-randomly (controlled for equal appearance of all locations) chosen to not allow extraction spatial patterns. One session consisted of 5 sample trials on 1 day and a test trial 24 hrs later. All locations (i.e., more stable locations in stable and overlapping) were counterbalanced across animals for each condition as well as within an animal across conditions to avoid general place preference effects. B. Panel: Exploration time. The total exploration time remained constant across conditions but a significant effect of trial was observed (P = 0.017). C. And D. Panel: DI (for statistical details see main text). The DI across sample and test trials showed a significant condition x trial interaction effect (P = 0.011). In the test trial, there was a significant condition effect (P = 0.008), and the DI was significantly above chance only for stable and overlapping condition (*P < 0.05). Left and right panel D same data. Data in S1 Data. DI, discrimination index.
Fig 3.
A. Panel: Trial structures for the 3 different conditions (object identity changed from trial to trial). In the overlapping condition, 1 location remains constant across all sample trials and the test trial, the second location varies. The locations in the last sample trial and in the test trial are equal. In the stable condition, the locations remain the same in all sample trials and 1 object is displaced in the test trial. In the random condition, the locations were pseudo-randomly chosen (controlled for equal appearance of all locations) to not allow extraction spatial patterns. One session consisted of 5 sample trials for 4 subsequent days and test trial 24 hrs later. All locations (i.e., more stable locations in stable and overlapping) were counterbalanced across animals for each condition as well as within an animal across conditions to avoid general place preference effects. B. Panel: Exploration time over the course of all 20 sample trials and test trial for each condition. Alternating white and grey shaded areas indicate the individual training days and test day. The total exploration time per sample trial remained constant across conditions; however, significant effects of trial, day, and a significant trialXday interaction were observed (condition P = 0.59; trial P < 0.001; day P < 0.001; trialXday P < 0.001). C. Panel: DI for all 20 sample trials and test trial across conditions. Alternating white and grey shaded areas indicate individual training days and the test day. D. Panel: DI per sample trial over the course of all 4 training days across conditions. A marginal significant effect for trial has been found (P = 0.09). More importantly, a significant conditionXtrial interaction was observed (P = 0.042), indicating only a build-up of preference for the less stable location over the daily trials in the overlapping but not stable or random condition. E. Panel: DI for each training day (the 5 sample trials for each training day averaged) and test day per condition. F. Panel: DI at the final training trial and test trial, which showed a significant trialXcondition interaction (P = 0.046). Memory performance was significantly above chance level in the overlapping condition for both the last sample trial and test trial (last sample P < 0.01; test P < 0.05). In the stable condition, only the test trial showed a significant effect (last sample P = 0.59; test P < 0.01). No significant effects were observed in the random condition (last sample P = 0.50; test P = 0.73). F and G same data. Data in S2 Data. DI, discrimination index.
Fig 4.
Object space task mouse: 4-week overlapping training.
A Panel: Trial structures for the 4-week version of the overlapping condition (object identity changed from trial to trial). Across the 4 weeks, 1 location remains constant across all sample trials and the test trial, the second location varies (5 trials per day, 5 days for the first 3 weeks). The first trial on Monday in week 2 (trial 26) as well as the final trial on Wednesday in week 4 (trial 76) function as 3-day and 5-day test trials, respectively. The stable location was counterbalanced across animals for each condition as well as within an animal across conditions to avoid general place preference effects. B Panel: Exploration time remained stable across the 3 weeks (P = 0.96). C Panel: The DI remains stable with preference for the less often shown location across the 3 weeks (P = 0.5). D Panel: Test to control for episodic memory effects the locations in the last sample trials and in the tests trial are equal. Both 3 days and 5 days after training, the animals showed a significant cumulative memory effect, with preference for the less often shown location (*P = 0.033, **P = 0.008). E. DI of each trial for the whole 4-week period. Data in S4 Data. DI, discrimination index.
Fig 5.
Object space task: Calcium imaging and electrophysiology.
Example of a mouse running the object space task with calcium imaging (Inscopix, A), example raw signal (B), and extracted calcium transients (C). D shows exploration times in implanted mice (Data in S6 Data) and E dwell time maps for a random set of 20 trial of unimplanted (left) and implanted (right) mice. F shows exploration times and G dwell times maps for rats (Data in S7 Data; G again random sample of 20 videos). Example of a rat (H) and mouse (I) running the object space task with electrophysiological implants, example raw LFP signal (J) and extracted unit activity (K). Further speed and position analysis of implanted and unimplanted animals: mouse implanted (speed 5.559 cm/s ± 2.323 cm/s, at border: 75.1% of frames, at center: 24.9% of frames. Data in S10 Data), mouse unimplanted (speed 7.497 cm/s ± 4.314cm/s, at border: 75.3% of frames, at center: 24.7% of frames. Data in S11 Data), rat implanted (speed 5.689 cm/s± 1.623 cm/s, at border: 69.1% of frames, at center: 30.9% of frames, Data in S12 Data), and rat unimplanted (speed 7.909 cm/s ± 3.629 cm/s, at border: 79.8% of frames, at center: 20.2% of frames. Data in S13 Data). n = 3 for rats, n = 6 for mice, random 10 trials were chosen for analysis.
Table 1.
Results of model fitting on rat behavioral data.
AIC and BIC both give model M1 as winner. All models fit the data significantly better than chance: pR2 score > 0. The table shows mean ± SEM of parameter values without outliers (|β|≥1,000).
Table 2.
Results of model fitting on mouse behavioral data.
AIC and BIC both give model M1 as winner. All models fit the data significantly better than chance: pR2 score > 0. The table shows mean ± SEM of parameter values without outliers (|β|≥1,000).
Fig 6.
Fit of computational model on data.
The optimized model fits the behavior of rats (A–B) and mice (C–D) better than chance. A. Histogram of goodness of fit (model likelihood) for all rats, averaged across conditions (first panel) and in the 3 experimental conditions separately. The vertical dashed line denotes average chance level. Note, however, that for each individual session, the model fits better than that expected by chance for that specific session. B. Optimal values for the model parameters α and β rat experiments. Each symbol corresponds to 1 experiment test (1 animal, 1 condition, conditions may be repeated across animals). For overlapping, the model requires an α≪1 in order to reliably accumulate evidence over a large number of trials. In contrast, for stable high values of α were obtained. The value of α for stable is nevertheless to a large extent immaterial for task performance because recent and remote trials contain the same information (the object configuration does not change). Double horizontal lines indicate y-axis discontinuity (to accommodate for outliers) C. and D. same as A and B for mice. D. Horizontal dashed lines indicate |β|≫20 in B and |β|≫10 in D. Data in S8 Data.
Fig 7.
The computational model can reproduce animal performance.
The model simulations reproduced the progressive build-up over trials of a significant DI in the overlapping condition for both rats and mice. It also reproduced a positive DI at the test trial for the stable condition. A. DI of model simulated with optimal parameter values (from Tables 1 and 2) for rats and mice. Triangles represent overlapping condition, squares stable, and grey circles random as in Fig 2. B. DI at test trial in model simulations. C. Time course of Entropy of Object 2 in simulations. The model predicts that high DIs are due to a high entropy (i.e., uncertainty) associated to Object 2 (moving object). D. Shows the DI at last training trial and test for different parameter values of α and β. Neophilic behavior (yellow) is obtained for positive β while neophobic behavior (blue) is obtained for negative β. Moreover, α < 0.5 is ideal to obtain a high DI in the overlapping condition, while higher α is acceptable for the test trial of the stable condition. Data in S9 Data. DI, discrimination index.