Article Figures & Data
Figures
- Figure 1.
Experimental schedules for the BM beginners and learners. Experimental schedules for the BM1 beginner, the BM3 learner and the BM1' learner (A), and those for the BM3 beginner and the BM1 learner (B). Each small circle represents a day. Each color of each circle represents an experimental phase of either BM experiment; blue, cyan, and green express the habituation, training, and probe test in the BM1 and BM1', respectively, while red, orange, and yellow are the habituation, training, and probe test in the BM3, respectively.
- Figure 2.
Architecture of the conventional 1-m diameter Barnes maze (BM1) and the 3-m diameter Barnes maze (BM3). A–F, The architecture of BM1 and BM3. Top, middle, and bottom rows indicate outlines (A, B), birds-eye views (C, D), and enlarged images of goals (E, F), respectively. In the outline, the display unit of vertical (y) and horizontal (x) axes in meters. The largest circle drawn by black solid line represents the edge of each maze. The “+” markers and blue-line circles around them represent the goal or dummy holes and the areas where hole visiting was scored, respectively. The radius of each area was set at ∼80 mm for the BM1 (A) and 270 mm for the BM3 (B). While these blue areas were designed so that these sizes in the BM3 were the triple of those in the BM1, the radius of the holes was ∼40 mm larger in the BM3 than in the BM1 (see Materials and Methods); therefore, we set the radius of the target fields in the BM3 to be the sum of the “margin” and exactly triple of the radius of the target field in the BM1. Namely, the number of errors was determined by the number that mice entered the blue-line circled areas until they goaled. Likewise, in the analysis of the probe test of spatial memory, time spent around each hole was the duration that mice stayed in the blue-line circled area. The black dashed-line circles represent the start areas with a lift transporting mice to the field. In strategy analysis, trajectory data while mice stayed in the start area since they entered the field were discarded, so that initial wobbly trajectory around the lift did not affect the classification of navigation strategy (see details in Materials and Methods). C, D, In the birds-eye views, the orange arrows point to the locations of distal spatial cues. The red and green squares represent escape box/tunnel and mice, respectively. Each number (#) in D corresponds to each component of the BM3 arena. Escape box and tunnel were zoomed in E, F, respectively.
- Figure 3.
Lower learning rate and inaccurate spatial representation in the BM3 compared with the BM1. A–C, Trial-based learning rate across training periods in conventional features in the BM1 and the BM3. Three trials per day were conducted for six successive days in the BM1 and 11 successive days in the BM3. The measured values of number of errors (A), and the normalized values of latency (B) and travel distance (C) to reach the goals were displayed. The horizontal axis and the vertical gray grid lines of each panel indicates training days and trials, respectively. Note that both latency and travel distance were normalized so that the values range between 0 and 1. Blue-colored and red-colored dots or lines represent the results of BM1 (n = 34) and BM3 (n = 56), respectively. Each dot represents the median value of each group in each trial. Solid lines represent learning curves estimated by nonlinear exponential curve fitting for the median of the groups. Box plots in the small insets in (A–C) represent distributions and comparisons of decay parameter β of the learning curves between the mouse group of BM1 and BM3. Asterisks indicate statistically significant differences. Learning curves fitted for raw values of latency and travel distance are shown in Extended Data Figure 3-2A,B. D–F, Daily basis learning curves across training periods in conventional features in the BM1 and the BM3. These scores were averaged over three trials per day. Blue and red solid lines represent changes in the median of measured values of number of errors (A) latency (B) and travel distance (C) of the BM1 and the BM3, respectively. Shaded areas indicate median absolute deviation within a mouse group each day. Asterisks indicate significant differences between the BM1 and the BM3. G, Strategy usage of the subjected mice across training days in the BM1 and the BM3. The left and right panel indicate the results of BM1 and the BM3. The vertical and horizontal axes are the proportion of the strategies and training days, respectively. In each stacked bar graph, blue, green, and yellow color represent random, serial and spatial strategy, respectively. Red asterisks indicate significantly different strategy usage between the BM1 and the BM3 in a given day. Samples of each strategy observed in the BM3 are shown under the stacked bar graphs. From left, random, serial spatial strategies are shown. Each colored line represents a trajectory classified as a strategy in a trial. These samples were chosen so that the sum of travel distances of samples within each strategy were comparable between the strategies. All trajectories were transformed so that the goal is located at the right top hole, noted as “Target.” The larger circles with black solid lines represent the edges of the BM3, while the smaller circles with black dashed lines represent the start areas. “+” markers represent hole locations. H, Time spent around each hole in the probe test. The horizontal axis indicates the locations of the holes expressed as angle differences from the target. The vertical and horizontal axes are search time for the individual hole and hole location indicated by angle from the target with 30° step, respectively. “Target” shown by a black dashed line is the goal hole for each mouse. Blue and red represent the results of BM1 (n = 34) and BM3 (n = 40), respectively. The solid lines are connected between the median values in each angle of each mouse group. The shaded areas are a range of median ± median absolute deviation of the data in each angle. Asterisks indicate significant differences between the BM1 and the BM3. We confirmed that these results were replicated even under durations such that the number of errors were statistically comparable between the BM1 and the BM3 (Extended Data Fig. 3-2C). The focal search patterns in scopolamine-treated mice, as reported in the BM1 (Suzuki and Imayoshi, 2017), were partially replicated in the BM3 (Extended Data Fig. 3-4).
- Figure 4.
Difference of network structures between the BM1 and the BM3. A–H, Changes of network features across training in the BM1 and the BM3. Temporal changes in the graphically displayed global networks in the BM1 and BM3 during spatial learning were illustrated in Extended Data Figure 4-1. The vertical and horizontal axes in each panel are calculated values of each network feature and training days. These scores were averaged over three trials per day. Blue and red represent the results of BM1 (n = 34) and BM3 (n = 56), respectively. Solid lines represent changes in the median value of each group in each day. Shaded areas represent error ranges between 25th and 75th percentile of the data within a group each day. Statistical results on each day are shown in Extended Data Figure 4-2. I–P, Network features in the probe test in the BM1 and the BM3. The vertical axis in each panel is each network measure. The blue and red represent the results of the BM1 and BM3 group, respectively. Each dot represents a mouse in either group. Squares represent median value in each group. Asterisks indicate significant differences between the BM1 and the BM3.
- Figure 5.
Effects of prior learning in the BM3 and BM1' on spatial navigation and learning in the subsequent BM1 task. A, Daily basis learning curves across training periods in conventional features in the BM1. These scores were averaged over three trials per day. From left, the measured values of number of errors, latency and travel distance are displayed. The mouse groups of BM3 learner (n = 16) and BM1' learner (n = 17), which experienced the BM3 and BM1' before the BM1, respectively, were compared with the Beginner (n = 34) group. The number of errors and travel distances of the BM3 learner and the BM1' learner were significantly lower than Beginner. Latency in the BM1' learner was significantly shorter than beginners. Asterisks indicate statistically significant differences. B, Daily strategy usage during BM1 training in the BM3, BM1' learner and Beginner mouse groups. Although usages of all navigation strategies were comparable between the BM3 and the BM1' learner in all training days, significant differences were detected in the comparison of BM3 learner and Beginner, and BM1' learner and Beginner. Red asterisks indicate significantly different strategy usages compared with the corresponding strategy usages in the beginner on a given day. C, Temporal changes of network features in the BM1 training of BM3, BM1' learner and Beginner. The vertical and horizontal axes in each panel are calculated values of each network feature and training days, respectively. These scores were averaged over three trials per day. Solid lines represent changes in the median value of each group in each day. Shaded areas represent error ranges between 25th and 75th percentile of the data within a group each day. Statistical results on each day are shown in Extended Data Figure 5-3. D, Time spent around each hole in the BM1 probe test of the BM3, BM1' learner and Beginner mouse group. The vertical and horizontal axis is time and hole location indicated by angle from the target with 30° step, respectively. “Target” expressed as a black dashed line is the goal hole for individual mice. Blue, red, and yellow represent the results of Beginner, BM3 learner and the BM1' learner, respectively. The solid lines are median values in each angle in each group. The areas are a range of median ± median absolute deviation of the data in each angle in each group. Asterisks indicate significant differences between the groups. E, Network features in the BM1 probe test compared between Beginner, BM3 learner and the BM1' learner. The vertical axis in each panel is each network measure. The blue, red, and yellow represent the results of Beginner, BM3 learner and the BM1' learner, respectively. Each dot represents a mouse in each group. Squares represent median value in each group. Asterisks indicate significant differences between the mouse groups. We confirmed that the learning curves in the BM1' were virtually similar to those in the BM1 (Extended Data Fig. 5-1). As expected, the facilitation effect from prior learning to the subsequent BM1 learning disappeared if environmental or task structures differed between prior learning and the BM1 learning (Extended Data Fig. 5-5). Not predicted by our model, facilitation effect would be limited, if prior BM learning was done in a smaller scale space than that in the subsequent BM learning (Extended Data Fig. 5-6).
Tables
Extended Data 1
Codes for data acquisition and analysis. Download Extended Data 1, ZIP file.
Extended Data Figure 3-1.
Summary for behavioral features. Download Figure 3-1, DOC file.
Extended Data Figure 3-2
Learning curves fitted for raw values of latency and travel distance and time spent around each hole given adjusted duration. While the BM1 and the BM3 would have different scale lower boundaries of latency and travel distance, respectively, while latency and travel distance should be compared between the BM1 and the BM3 within the same scale. So, latency and travel distance were normalized across trials so that the values range between 0 and 1 (Fig. 3B,C). A, A learning curve for raw values of latency. The vertical and horizontal axis indicates latency in second and training days, respectively. The same curve fitting method was used as Figure 3B. B, A learning curve for the raw values of travel distance. The vertical and horizontal axis indicates travel distance in centimeters and training days, respectively. The same curve fitting method was used as Figure 3C. The small inset in panels A and B is the distribution of estimated decay parameter β in the BM1 and the BM3. Asterisk indicates significant differences between the BM1 and the BM3. C, Time spent around each hole in the probe test under adjusted duration for the BM1 and the BM3. Duration for the BM1 and the BM3 was 90 s and 150 s, respectively, and was balanced so that the number of errors were statistically comparable between the BM1 and the BM3. Other than that, the graph format is identical to Figure 3H. Download Figure 3-2, TIF file.
Extended Data Figure 3-3
Statistical results of strategy analysis for the BM1 and the BM3 in the training phase. Download Figure 3-3, DOCX file.
Extended Data Figure 3-4
Effects of scopolamine administration on the BM3 probe test. Time spent around each hole in the probe test. Asterisks indicate significant differences between the two groups. Red and blue represent the scopolamine-treated (SCOP) and untreated (NO-SCOP) mouse groups, respectively; the former is cohort S1 (n = 16), while the latter is the pool of Cohort 2 (n = 20) and the first instance of Cohort 3 (n = 20) in Table 1.. A mixed-design 2-way [Scopolamine (SCOP, NO-SCOP) × Hole (1∼12)] ANOVA for the time spent around each hole detected significant interaction between the scopolamine and hole, F(11,594) = 2.40, p = 0.01, ηp2 = 0.04. Multiple comparison testing detected that the time spent around target +30° was significantly longer in the SCOP than in the NO-SCOP whilst time spent around the opposite hole from the target was significantly shorter in the SCOP than in the NO-SCOP. Because a dummy escape tunnel was attached under the hole opposite to the goal hole (see Material and Methods), the NO-SCOP but not SCOP mice might search around it, once they observed that the true escape tunnel no longer exists at the goal hole. Download Figure 3-4, TIF file.
Extended Data Figure 4-1
Visualization of temporal changes in the exploration networks during spatial learning in the BM1 and BM3. A, Temporal changes in the global exploration networks in the BM1 spatial learning. Network structures of exploratory behaviors formed by the dynamic node generation method are plotted (Suzuki & Imayoshi, 2017). Small colored dots and light gray lines represent nodes and links in each mouse (local network). The nodes are located on Cartesian coordinates of the BM1. Nodes are color-coded depending on their polar coordinates. To visualize the global network structure, all local networks of the mice on a single training day were projected on a single plane. Colored larger circles and dark gray lines are global nodes and links in global networks, respectively. Local nodes belong to any one of the global nodes. Likewise, a set of local links are summarized as a global link. The size of a global node is based on log-transformation of the number of nodes that belong to the global node. Likewise, the thickness of a global link is log-transformation of the number of links that belong. B, Topological expression of BM1 global networks. Global nodes were sorted by rank-order of degree and plotted on polar coordinates, so that the node with the highest degree was located at 0 degrees while the lowest one was located at 360 degrees. Circles and lines represent global nodes and links, respectively. Global nodes were ranked to any one of 30 ranks depending on its degree within each group. Then, they were colored according to the rank; the rank first node has a lot of links and is colored by white, and rank 30th node has few links and is colored by black. C, D, Temporal changes in the global networks and their topological expressions of the BM3 spatial learning were displayed. Download Figure 4-1, TIF file.
Extended Data Figure 4-2
Statistical results of network analysis for the BM1 and the BM3 in the training phase. Download Figure 4-2, DOCX file.
Extended Data Figure 4-3
Statistical results of network analysis for the BM1 and the BM3 in the BM1 probe test. Download Figure 4-3, DOCX file.
Extended Data Figure 5-1
The experiment setup and spatial learning curves in BM1 and BM1'. A, Spatial cues used in the BM1 (left panel) and the BM1' (right panel) were indicated with orange arrows. Note that all setups other than the cues were identical between the two mazes. B, Daily basis learning curves across training periods in conventional features in the BM1 and BM1'. These scores were averaged over 3 trials per day. From left, the measured values of number of errors, latency and travel distance are displayed. Only in a mixed-design 2-way [Spatial cues (BM1, BM1') × Day (1–6)] ANOVA for travel distance, the main effect of spatial cues was significant, F(1,49) = 6.72, p = 0.01, ηp2 = 0.12. Asterisks indicate significant differences between the two groups. Download Figure 5-1, TIF file.
Extended Data Figure 5-2
Statistical results of strategy analysis for the BM1’ learner, the BM3 learner, and the Beginner in the BM1 training. Download Figure 5-2, DOCX file.
Extended Data Figure 5-3
Statistical results of network analysis for the BM1’ learner, the BM3 learner, and the Beginner in the BM1 training. Download Figure 5-3, DOCX file.
Extended Data Figure 5-4
Statistical results of network analysis for the BM1’ learner, the BM3 learner, and the Beginner in the BM1 probe test. Download Figure 5-4, DOCX file.
Extended Data Figure 5-5
Prior fear-conditioning experience does not impact on the subsequent BM1 spatial learning. Daily basis learning curves of the CFC learner n = 8) and Beginner (n = 34) across training periods in conventional features in the BM1. These scores were averaged over 3 trials per day. From left, the measured values of number of errors, latency and travel distance are displayed. Mixed-design 2-way [Instance (CFC learner, Beginner) × Day (1–6)] ANOVA detected neither main effect of Instance nor interaction between instance and day for all features. Download Figure 5-5, TIF file.
Extended Data Figure 5-6
Limited effects of spatial navigation and learning in the subsequent BM3 task by prior learnings in the BM1. A, Daily basis learning curves across training periods in conventional features in the BM3. These scores were averaged over 3 trials per day. From left, the measured values of number of errors, latency and travel distance are displayed. The mouse group of the BM1 learner (n = 13), which experienced the BM1 before the BM3, was compared with the Beginner (n = 56) group. Although the significant changes were detected in latency and travel distances, no significant difference was detected in the number of errors. B, Strategy usage across training days in the BM3 and comparison between BM1 learner and Beginner. Limited changes were detected only at the Day 4 results. C, Temporal changes of network features in the BM3 training of BM1 leaner and Beginner. Statistical results on each day are shown in Extended Data Figure 5-8. D, Time spent around each hole in the BM3 probe test. Only significant main effect of instance was observed; exploration time in the BM1 learner was significantly longer than that in the Beginner, regardless of hole location. E, Network features in the BM3 probe test compared between BM1 learner and Beginner. Asterisks indicate significant differences between the two groups. Download Figure 5-6, TIF file.
Extended Data Figure 5-7
Statistical results of strategy analysis for the BM1 learner and the Beginner in the BM3 training phase. Download Figure 5-7, DOCX file.
Extended Data Figure 5-8
Statistical results of network analysis for the BM1 learner and the Beginner in the BM3 training phase. Download Figure 5-8, DOCX file.
Extended Data Figure 5-9
Statistical results of network analysis for the BM1 learner and the Beginner in the BM3 probe test. Download Figure 5-9, DOCX file.
In this issue
