Article Figures & Data
Figures
- Figure 1.
Task design, linear translation of pool, and probe test conditions. a, The schematic illustrates the two locations in the testing room where the pool (large black circles, West and East) was positioned during hidden platform training. Respective platform locations, A–C. Mice were trained in the west pool position to platform A or B, and in the east pool position to platform B or C. Gray arrowheads indicate the points where the mice were released into the west pool during training trials. Extramaze visual cues were positioned on a circular black floor-to-ceiling curtain (2.4 m diameter) that surrounded the pool. The vertical break in the curtain at approximately two o'clock represents the point of entry through the curtain. b, Before a probe test, the pool was translated linearly from the west position (thin dashed circle) in the testing room to the east position (solid black circle), or vice-versa (schematic not shown). Gray arrowheads indicate the two points (north and south) where the mice were released during the probe test. For mice trained to platform location B (relative vs absolute condition), the translation permitted examining preference for mice to exhibit place or absolute responding [swimming to the absolute search (dashed circle, A) zone during the probe] or relative responding [swimming to the relative search zone (dashed circle, R) during the probe test]. For mice trained to platform location A (relative-only condition), the translation resulted in the trained place being inaccessible during the probe test. Behavior in this condition was characterized by swimming to the relative search zone or to the opposite search zone. Search behavior was quantified by a heading error score for each probe test, which was computed by plotting the initial swim path from release to arrival into the relative search zone or relative/opposite search zone and measuring the angular deviation from a direct, straight path to the relative search zone (dashed lines). c, For the rotation of visual cues and rotation of visual cues plus translation of the pool conditions, the extramaze cues were rotated 90° clockwise before the probe test in the translated pool (no pool translation in the former condition). The entry point through the curtain, the two release points, and the search zones of interest, R′ and A′ were also rotated 90° clockwise.
- Figure 2.
Absolute navigation training and performance of mice under the three probe test conditions (no shift, relative vs absolute, and relative only). a, b, The three groups acquired the task equivalently as illustrated by trial block plots of escape latency (a) and cumulative distance to platform center (b). The arrow indicates when the probe test was imposed. c, No-shift (open circle) mice exhibited a strong preference for searching for the platform in the trained quadrant during the probe test (*p < 0.01 vs other quadrants), indicating that all mice acquired the absolute memory for the platform location. d, Mice of the relative versus absolute (closed circle) and relative-only (gray triangle) conditions swam first to the relative search zone during the probe test in the translated pool instead of the absolute/opposite search zone. e, The mice also exhibited a significantly shorter latency to first enter the relative search zone compared to the absolute/opposite search zone (αp < 0.001; βp < 0.001). f, The polar plot of the distribution of individual heading error measures (black circles, relative vs absolute condition; gray triangles, relative-only condition) indicates that the majority of error scores were clustered at ∼10°, corresponding to relative responding, and a minor cluster of error scores at ∼60°, corresponding to absolute responding. The arrow denotes the combined mean vector angle of 10.3° for both conditions. The length of the arrow represents the mean vector length (r = 0.86) that denotes the variability in heading error scores, with no variability indicated by r = 1. g, Representative initial swim paths are depicted for mice of the no-shift (top), relative versus absolute (middle), and relative-only (bottom) conditions. For the no-shift mice, the location of the trained platform A, B, or C is depicted; the relative position of the platform during the training trials for these mice is indicated by the small dashed lined circle inside the larger dashed circles that indicate the respective search zones. For the relative versus absolute and relative-only mice, the respective search zones are indicated by R (relative), A (absolute), and O (opposite). Error bars indicate SEM.
- Figure 3.
Evidence of absolute and relative responding in mice at an early stage of MWM training. a, Two cohorts of naive male C57BL/6J mice were trained in the MWM for eight trials (Day 2; n = 12) or 16 trials (Day 4; n = 12) in the west position of the testing room. Just before a 30 s probe test after the 8th or 16th training trial (arrows), the pool was translated to the east position in the room, and mice were released at the north or south point. The Day 2 mice failed to exhibit a preference for relative (n = 6) or absolute responding (n = 6) during the probe after the eighth trial. In contrast, the Day 4 mice predominantly swam first to the relative search zone (n = 8) as opposed to the absolute search zone (n = 4). b, The mice probed after 2 d of training exhibited a weak preference for the relative zone over the absolute zone. Interestingly, this preference for relative responding was more pronounced in the mice that received the probe test after 4 d of training. c, The polar plot of the distribution of individual heading error measures (deviations from a direct path to the relative search zone) indicates that scores were clustered at ∼0°, indicating relative responding. The mean heading error for Day 2 mice probed after 2 d of training was 19.8°, indicated by the heading error vector (black arrow), while the Day 4 mice had a mean heading error of 14.3°, as indicated by the heading error vector (gray arrow). These data indicate that relative responding in the MWM develops progressively over an early stage of training. d, A second naive cohort of 12 male C57BL/6J mice was trained in the MWM for 16 trials in the west position of the testing room. The extramaze cues were rotated 90° clockwise just before a 30 s probe test, but the pool remained in the west. The initial trajectory that each mouse took from the release point and the location where the mice searched for the platform both shifted by ∼90°. The mice exhibited a strong preference for searching in the A′ zone (n = 9) over the O′ zone (n = 3) in the presence of the rotated cues. e, The polar plot of the distribution of individual heading error measures (deviations from a direct path to the A′ search zone) indicates that the error scores shifted by ∼90°. The mean heading error was −105.5°, as indicated by the heading error vector (arrow). These data indicate that the distal cues exert stimulus control over the platform search behavior of male C57BL/6J mice, suggesting that the mice engaged in platform search in the shifted absolute location. Error bars indicate SEM.
- Figure 4.
Extramaze cues and internal cues influence relative responding of mice during the probe test in the translated pool. a, A naive cohort of 12 male C57BL/6J mice was trained in the MWM for 28 trials in the west position of the testing room. The pool was translated linearly to the east position of the room, but here the extramaze cues were rotated 90° clockwise just before the probe test, denoted by the arrow at trial 28 (Fig. 1c). Mice again exhibited a strong preference for relative responding over absolute responding in the presence of the rotated cues. b, The polar plot of the distribution of individual heading error measures indicates that the error scores shifted by ∼90° compared to the control scores exhibited in Figure 2f. The mean heading error was −81.6°, as indicated by the heading error vector (length of the heading error vector, r = 0.93). c, The polar plot of the distribution of individual heading error measures indicates that the error scores were quite distributed from approximately −30 to +80° compared to the control scores exhibited in Figure 2f. The mean heading error was 25.4°, as indicated by the heading error vector (length of the heading error vector, r = 0.86). d, Representative swim paths for four mice depict the direct swim to the R′ search zone over the A′ zone. These data indicate that the extramaze cues exert stimulus control over relative responding. The mice were retrained in the pool in the west position for eight trials, and then the pool was again translated linearly to the east position for another probe test (a, arrow at trial 36). Before the probe test, each mouse was placed into a closed cardboard box and disoriented by gently rotating the box clockwise and counterclockwise while it was carried around the MWM for 60 s. After a 30 s rest, the box was carried to the release point at the north or south of the pool, and the mouse was released into the pool for a 30 s probe test. Disorientation eliminated the preference of the mice for relative responding, with the mice exhibiting essentially equal preference for relative and absolute responding. e, Representative swim paths for four mice demonstrate that disorientation led mice to swim in circuitous paths to the respective search zones. Error bars indicate SEM.
- Figure 5.
Histological verification of bilateral infusion cannulae placement into the anterior thalamic nuclei. a, Representative photomicrographs depicting infusion cannula track through the ATN in the left hemisphere (left) and right hemisphere (right). The anterodorsal thalamic nuclei (AD), anteroventral thalamic nuclei (AV), anteromedial thalamic nuclei (AM), fimbria (fi), stria medullaris (sm), and the third ventricle (3V) are outlined or identified in white. Arrowheads indicate the track of the infusion cannula. b, The individual placements of infusion sites (filled black circles) for the 14 mice are presented against the atlas plates from −0.58 to −0.94 mm from bregma (Franklin and Paxinos, 2008).
- Figure 6.
Inactivation of the anterior thalamic nuclei abolishes relative responding in the water maze. a, A naive cohort of 14 male C57BL/6J mice were trained for 28 trials in the water maze in the west position of the testing room. Each mouse received a bilateral microinfusion of aCSF or muscimol (2 μg/0.25 μl/side) 20 min before the probe test in the translated pool. Mice were trained for eight more trials over the following 2 d and then received a second bilateral microinfusion of aCSF or muscimol with treatment assignments reversed from those of the first microinfusion. Twenty minutes after each microinfusion, each mouse received a 30 s probe test (arrows) in the translated pool. b, The mice exhibited a strong preference for relative responding over absolute responding after intra-anterodorsal thalamic aCSF, consistent with results shown in Figure 2. In contrast, the mice exhibited an overwhelmingly preference for absolute responding over relative responding after intra-anterodorsal thalamic muscimol. c, After aCSF microinfusion, the mice exhibited significantly shorter latencies to the relative search zone than the absolute search zone (αp < 0.001). After muscimol microinfusion, the mice instead exhibited significantly shorter latencies to the absolute search zone than the relative search zone (βp < 0.001). d, The polar plot of the distribution of individual heading error scores for the mice reveal that scores for aCSF treatment were clustered at ∼0° (mean heading error, 1.2°, black arrow; length of the heading error vector, r = 0.96), while those of muscimol treatment were clustered at ∼60° (mean, 54.8°, gray arrow; length of the heading error vector, r = 0.96), consistent with absolute responding. e, Representative swim paths for five of the mice indicate that after intra-anterodorsal thalamic aCSF (top), swim paths were direct to the relative search zone, while after intra-anterodorsal thalamic muscimol, the same mice (bottom) swam to the absolute search zone via a much less direct and more circuitous paths. Error bars indicate SEM.
- Figure 7.
Histological verification of bilateral infusion cannulae placement into the CA1 region of the dorsal hippocampus. a, A representative photomicrograph depicting the infusion cannula track through the right dorsal CA1 region (right). The left panel depicts the left dorsal CA1 region for comparison purposes, from the mouse brain atlas of Franklin and Paxinos (2008). b, The individual placements of infusion sites (black symbols) for the 11 mice are presented against the atlas plates from −1.70 to −2.46 mm from bregma.
- Figure 8.
Relative responding is not affected by inactivation of the dorsal CA1 region of the hippocampus. This experiment was conducted identically to that presented in Figure 6, except that mice received bilateral microinfusion of muscimol or aCSF into the CA1 region of dorsal hippocampus before a 30 s probe test in the translated pool. a, Eleven naive male C57BL/6J mice were trained for 28 trials in the water maze in the west position of the testing room. Twenty minutes after microinfusion of aCSF or muscimol (2 μg/0.25 μl/side), each mouse received a 30 s probe test in the translated pool (arrows). b, The mice exhibited a strong preference for relative responding over absolute responding after intra-CA1 aCSF, consistent with results shown in Figure 2. Interestingly, relative responding was also observed after intra-CA1 muscimol. c, After intra-CA1 aCSF or muscimol, mice exhibited significantly shorter latencies to the relative search zone than the absolute search zone (α,βp < 0.001). d, The polar plot of the distribution of individual heading error scores for the mice shows clustering at ∼10° (aCSF mean heading error vector, 10.4°, black arrow, r = 0.89; muscimol mean heading error vector, 11.4°, gray arrow, r = 0.87), indicative of relative responding. e, Representative swim paths for five of the mice indicate that after intra-CA1 infusion of aCSF (top) or muscimol (bottom), swim paths were direct to the relative search zone. Error bars indicate SEM.
- Figure 9.
Relative responding is impaired after intra-anterior thalamic FCM, but spared after intra-CA1 FCM. Naive C57BL/6J mice (n = 9) were implanted with bilateral cannulae directed at the anterodorsal thalamic nuclei. The mice were then trained for 28 trials in the water maze in the west position to find the platform at position B, as described for the other experiments. Twenty minutes before a 30 s probe test in the translated pool, the mice received bilateral microinfusion of FCM or PBS vehicle. After 2 d of retraining in the west, the mice received a second microinfusion when treatment assignments were reversed. a, Epifluorescent image acquired at 4× magnification illustrating the distribution of FCM within the left ATN from a representative 50 μm coronal section. The fluorophore can be seen within the anterodorsal nuclei (AD; outlined with dashed white line), but also appears to have spread into the anteroventral nuclei (AV) and anteromedial nuclei (AM) as well. 3V, Third ventricle; sm, stria medullaris. b, Epifluorescent image illustrating the distribution of FCM (red) after intrathalamic microinfusion against green fluorescent background staining for Nissl (NeuroTrace; Invitrogen). The dashed white line depicts the boundary of the anterodorsal thalamic nuclei. c, Microinfusion of FCM into the ATN abolished relative responding observed after intrathalamic PBS. Intrathalamic PBS-treated mice swam first to the relative search zone during the probe test in the translated pool (αp < 0.01). In contrast, intrathalamic FCM-treated mice swam first to the absolute search zone, consistent with absolute responding (βp < 0.01). d, e, Epifluorescent images from two cases acquired at 4× magnification, illustrating the distribution of FCM within representative 50 μm coronal sections through the dorsal hippocampus after bilateral microinfusion into the CA1 region. The fluorophore appears to be distributed within the dorsal hippocampus with some spread ventrally to the dentate gyrus. f, Microinfusion of FCM into the dorsal CA1 did not affect relative responding of mice. Intra-CA1 PBS-treated mice swam first to the relative search zone, as did the same mice after intra-CA1 FCM (αp < 0.01; βp < 0.04). Error bars indicate SEM.
