Conceived and designed the experiments: AD HS IG. Performed the experiments: AD. Analyzed the data: AD. Contributed reagents/materials/analysis tools: AD HS. Wrote the paper: AD IG. Contributed to writing the paper: HS.
The authors have declared that no competing interests exist.
When introduced into a novel environment, mammals establish in it a preferred place marked by the highest number of visits and highest cumulative time spent in it. Examination of exploratory behavior in reference to this “home base” highlights important features of its organization. It might therefore be fruitful to search for other types of marked places in mouse exploratory behavior and examine their influence on overall behavior.
Examination of path curvatures of mice exploring a large empty arena revealed the presence of circumscribed locales marked by the performance of tortuous paths full of twists and turns. We term these places
In an environment devoid of proximal cues mice mark a locale associated with arousal by twisting and turning in it. This creates a self-generated, often centrally located landmark. The tortuosity of the path traced during the behavior implies almost concurrent multiple views of the environment. Knot-scribbling could therefore function as a way to obtain an overview of the entire environment, allowing re-calibration of the mouse's locale map and compass directions. The rich vestibular input generated by scribbling could improve the interpretation of the visual scene.
Exploration is a central component of human and animal behavior that has been studied in rodents for almost a century. It is presently one of the main models for studying the interface between behavior, genetics, drugs, and the brain. Until recently the exploration of an open field by rodents has been considered to be largely stochastic. Lately, this behavior is being gradually deciphered, revealing reference places called home bases, from which the animals perform roundtrips into the environment, tracing well-trodden paths whose features contribute to our understanding of navigation, locational memory, cognition-, and emotion-related behavior. Using advanced computational tools we discover so-called
When introduced into a novel environment rats establish in it a preferred place characterized by the highest frequency of visits, by the highest cumulative dwell time, by an upper bound on the number of stops per roundtrip performed from it, by low outbound trajectory speed and high inbound trajectory speed, with the speed relationship reversed in later stages of the session
In contrast to the situation with rats, there is an ambiguity regarding the establishment of home bases in mice: while some studies mention home bases, their establishment is mostly of a low occurrence in the forced exploration setup. Even though mice readily establish home bases near physical objects
When mice used as a control group in another study were injected with saline and placed in the exposed portion of a large open field arena, they established in it preferred places, typically not more than one per session, which they visited sporadically, tracing in them a tortuous path full of twists, turns, and bends that looked like a knot. We termed these places
To uncover knots we utilize a basic feature of the locomotor path – its curvature. Depending on the size of the window that is used for measuring path curvature this measure discloses various features of path texture, from overall large-scale curvature to fine-grained tortuosity
Observations of saline-injected mice revealed that when mice were released into the arena at a distance from the wall they showed a preference for a selected place often at the location of release, but also at other locations. The phenomenon is illustrated in the behavior of 3 mice, two of whom established a preferred place around the location of their release, and one that established such a place near the wall (
The mice were released at the same location, at the 12 o'clock position 50 cm away from the wall. Dark patches reflect high path density.
The leftmost figure in each horizontal line of figures shows boundaries of a preferred place within the arena; the rest of the figures in each horizontal line represent close-ups of 3 separate entries into that place. Red line presents the boundary of the knot, computed across the whole session, and black line presents the path. Blue arrows indicate the entry and exit paths into and out of the knot.
To capture the knots and map their boundaries we produced visual representations that highlighted places with knots by computing momentary path curvatures for all locations in the arena, and obtained a contour plot that mapped their boundaries (see
The colors represent path curvature measured with a 20cm window. Locations without paths and locations with low path curvatures are shown in blue, locations with medium curvature are shown in green, and locations with high path curvature are shown in red.
A matrix whose cells present these values of curvature per each square in the arena portrayed the level of curvature across the arena topography. To highlight areas characterized by similar curvature we used a spatial smoothing algorithm with a Gaussian kernel (see
To map the boundaries of knots we developed an image processing algorithm identifying areas filled with the red color that signified high curvature patches, and then filtered out the patches of u-turns along the wall (
A. A selected mouse-session cumulative path plot. B. A contour plot of fine-grained curvature topography of the same session. C. High curvature patches are circumscribed by a black boundary. D. The boundaries of u-turns are drawn in black. These patches of u-turns are filtered out. The boundary of the knot is singled out and drawn in red. In both B and C the colors represent path curvature measured with a 20cm window. Curvature values are represented by the same colors as in
A high density of paths does not imply the presence of a knot.
A. The path plot of a mouse-session portrays a circular bundle of paths. B. The contour plot of the same circular bundle of paths does not pick up the high path density patches because path curvature is of a different scale. Curvature window size and colors are as in
Knots were established at the location of the release of the mouse into the open field in 25 out of 41 mouse-sessions, in which knots were established. Mice left the release location almost immediately after being placed in it (3.8–34.3 sec), implying that the subsequent selection of this place for knot-scribbling did not depend on a long initial stay in it. The knots were visited for short periods of time (1.1–5.4 sec): unlike home bases, these places are not characterized by long visits. As illustrated in
Each horizontal line represents the visits of a selected mouse. Regular visits, not involving scribbling, are shown in gray and knot-scribbling visits are shown in red. Vertical bars' widths represent the durations of the visits.
A. Dark patches represent places with high path curvatures (knots and u-turns) established by a selected mouse across 10 successive sessions. The patches are shown in gray and the arena in cyan. Sessions are numbered. B. A superposition of all the patches, representing all knot and u-turn locations in the arena during 10 successive sessions, per mouse, for the 8 mice. Each circle represents 10 successive mouse sessions with knots superimposed across 10 sessions for each tested mouse. The shades of gray of the patches represent the number of superimposed patches, which in turn represent the number of knots established in the same location across the 10 session experiment: the darker the patch the higher the number of knots established in that location. The indices in the right upper corner indicate the number of sessions during which the mouse established a knot as a proportion of the total number of recorded sessions. One mouse had only 9 recorded sessions because of a technical failure.
To see whether home base locations coincide with knot locations we inspected the places characterized by two home base properties - the highest number of stops and the highest cumulative dwell time (see
A. The degree of overlap between these 3 types of locations in the 3 selected mice. B. The degree of overlap between the 3 types of locations in all mice. For each mouse, the occurrence of every type of overlap was calculated as the percentage of the total number of sessions in which that mouse established a knot. Each row represents the types and percentages of overlap for each of the mice.
To examine whether knots were characterized, like home bases
Ratio values per visit are presented by black dots; fitted linear regression is presented by a gray line.
Since there was no change in the ratios (
A. Rate of change of the outbound/inbound speeds' ratio within session as a function of excursion number. B. Median value of the outbound/inbound speed ratios per session.
To examine whether uninjected mice also established knots, and whether the location of introduction influenced knot frequency, we performed a subsequent experiment using intact mice that were either placed near the wall or in the center. The center-placing location was the same for the intact and for the injected mice (see
A. Knots established between 12 o'clock positions and center in three intact mice placed in the center area. B. Knots at about 8 o'clock positions away from wall in three intact mice placed near the wall at the 12 o'clock position.
Saline injections (Center) | No injections (Center) | No injections (Wall) | |
Number of mice | 8 | 9 | 9 |
Number of sessions | 10 | 7 | 7 |
Total number of analyzed sessions | 79 | 63 | 63 |
Number of knots established at the place of the introduction into the arena | 25 (60.9%) | 9 (56.3%) | 0 (0%) |
Total number of knots | 41 (51.9%) | 16 (25.4%) | 8 (12.7%) |
Different mice showing knots | 8 (100%) | 7 (77.8%) | 5 (55.6%) |
Percentages of the established knots (row 5) are calculated as a percentage out of the total number of mouse-sessions in this study (row 3). Percentages of the knots established at the place of the introduction (row 4) are calculated as a percentage out of the total number of knots (row 5).
In this study we uncover a new type of preferred places called knots, established by mice during exploratory behavior of an open field arena. The behavior generating these places, knot-scribbling, is performed by the mice during sporadic recurrent visits to the knot during the session. Knots are also characterized by a high path density, by frequent visits that do not involve scribbling, and by a higher speed on the way into the knot than on the way out of it. There is at most one knot per session and it is established in a circumscribed location having distinct boundaries. Knots are often established across successive sessions in the same place, but also in two or even three different places. They can be isolated solely on the basis of the curvature of the paths included in them. To make them “float” out of the background of paths we had to design and use an algorithm that calculated path curvature at the fine-grained scale that characterizes them. The algorithm scans the whole arena, highlighting locations with high cumulative path curvature and identifying their boundaries.
Unlike home bases, knots are neither characterized by maximal cumulative dwell time, nor by maximal number of stops. The reversal of the outbound-inbound speed ratio characterizing rat home base behavior across sessions
Mice tended to perform knot-scribbling at the point of introduction into the arena although they left these places immediately after being placed there. A preference for the point of entry, expressed by a longer-than-expected dwell time spent in the entry quadrant has been described in rats
Deciphering the function of knots will require extensive experimental manipulations, not available in this, mostly descriptive, study; some hypotheses may, however, be appropriate: having established a place of significance in an otherwise empty arena the mouse marks this place with twists, bends and turns that provide it with multiple, almost simultaneous, views of the explored environment, all from a single reference place. Perhaps while scribbling at the knot the mouse recalibrates its locale map and compass directions, in much the same way as a robot designed by Steinhage and Schoner
It has recently been demonstrated experimentally that a short vestibular stimulus can influence and even improve the perception and subsequent reconstruction of a much longer lasting visual stimulus
The structure of the knot, its effect on the behavior of the mouse when away from it, and the context of its formation, all rule out the possibility that the knot replaces a home base, or is interchangeable with a home base, or fulfills a similar function to a home base. A recent study reveals that when mice are exposed to a setup including a shelter and a free access to a large open field arena they demonstrate the full blown home base phenomenon
Saline-injected mice: 8 experimentally naive 8–11 week old male C57BL/6J mice (Charles-River, Canada) were used.
Intact mice: 18 experimentally naive 8–11 week old male C57BL/6J mice (Charles-River, Canada) were used.
All mice were individually housed in a temperature controlled colony room (22°C) under a 12h light-dark cycle, with free access to food and water. Mice were allowed to acclimatize to the colony room for two weeks following arrival and were handled two minutes daily for 7 days before the start of the experiment. All treatment and testing was conducted during the light hours.
Animals were housed and tested in compliance with the guidelines described in the Guide to the Care and Use of Experimental Animals (Canadian Council on Animal Care, 1993).
The large open field was a 220cm diameter circular arena with a non-porous gray floor and a 50cm high, primer gray painted, continuous wall. Several landmarks could be seen near the walls of the room where the arena was located. In particular, there was a table and a stand on which the lights were fixed. The arena was recorded using a resolution of 30 samples per second. The animal's movement was tracked using Noldus EthoVision automated tracking system
Saline-injected mice: On the day of testing, animals were weighed, transported in their home cages to an adjoining non-colony testing room, and administered a sub-cutaneous saline injection. Immediately afterwards, mice were placed within a small opaque box 50 cm from the wall of the circular open field arena at the 12 o'clock position. After 20 seconds the box was lifted, and a 55-min session began. Each mouse was videotaped for 10 sessions (2 sessions a week). Animals were tested during the light phase of the diurnal cycle.
Intact mice: Same as above except that the stage of injection was skipped, and mice were placed either near the wall, at the 12 o'clock position (n = 9), or 50 cm away from the wall at the 12 o'clock position (n = 9). The experiment with the intact mice was carried out in the same setup, after the completion of the experiment with the saline-injected mice.
The raw data obtained from the tracking system were smoothed using a specialized algorithm implemented in the stand-alone program “SEE Path Smoother”
A measure of curvature was computed for all data points obtained from the session. Curvature was measured as described previously (
Example of the computation of the curvature of the path at data point B. The curvature in degree/cm is defined as θ/(2
The algorithm for defining the knot contours was as follows:
Curvature was computed for all data points obtained from the session.
The arena was divided into 5×5cm squares (see
In order to highlight areas characterized by similar curvatures, we used a spatial smoothing algorithm with a Gaussian kernel (MatLab R2007a). The obtained matrix containing smoothed values of curvatures was used to create a 2D contour plot, and then the plot's contour lines were used in order to find the areas characterized by high path curvatures.
Examination of the paths traversed by the mice during the session revealed that some of the patches with high path curvature demarcated u-turns performed by the mice in close proximity to the wall. Since u-turn locations did not exceed a distance of more than 30 cm from the wall, knots were defined as patches of high curvature whose boundary extended for more than 30 cm away from the wall (see for example rightmost session in
To examine whether the establishment of a knot at the location where the mouse was placed was a result of a prolonged initial stay in it we measured the delay until the first exit from that place. Therefore, only sessions during which mice established a knot at that location of entry were used in the analysis. A median value of the duration of the delay until the first exit from a place later established as a knot was computed for each mouse across all sessions during which it indeed established a knot.
For each preferred place a circle of minimal area circumscribing it was computed. Maximal inbound speed was computed as a quantile 99 of speeds exhibited along the inbound path located within a 10cm ring situated around the circumscribing circle. Maximal outbound speed was computed similarly for the outbound path. Then the maximal outbound/inbound speeds' ratios for each excursion were computed. The dependency of maximal outbound/inbound speed's ratios (
Contour plots highlighting path curvature in the sessions of all saline-injected mice. Each horizontal line represents, from left to right, all the sessions of a mouse. The colors represent path curvature measured with a 20cm window. Locations without paths and locations with low path curvatures are colored in blue, medium curvature locations are colored in green, and high curvature locations are colored in red. Red high curvature patches along and near the wall reflect the prevalence of u-turns performed by the mice along the arena boundary. As shown, knots are established mainly at the location of the introduction into the arena (6 mice, 25 sessions, at about 12 o'clock position at a distance of 50 cm from wall), but also in other locations (near center toward 5 o'clock: mouse #1, session 1; #2, session 1, #3 session 2, #4 sessions 3,4,5,8; #8 sessions 2,3. near wall at 11 o'clock: mouse #2 session 4. Between center and 9 o'clock: mouse #7 session 10.) Some mice use only the place of entry for knot establishment (mouse #5) whereas others alternate between two (mice #1, #3, #8), or even 3 locations (mouse # 2). Every mouse established a knot during at least one session. The last session of the 5th mouse was not recorded because of a technical failure.
(8.29 MB TIF)
Timing and frequency of visits to knots in all saline-injected mice. Each plot shows the timing of visits to a knot across a single mouse- session. Visits to the knots are marked by vertical bars whose width is proportional to the visit's duration. Visits containing knot-scribbling are shown in red and all other visits in gray. As can be seen from the distribution of the red lines, knots are visited sporadically and frequently and visits include knot-scribbling throughout the session.
(2.31 MB TIF)
Contour plots representing the number of visits per location in all saline-injected mice. Each horizontal line represents, from left to right, all the sessions of a mouse. The colors represent the number of visits to locations (as defined by a 5×5cm grid). Locations with a low number of visits or no visits are colored in blue, locations with a medium number of visits are colored in green, and locations with a high number of visits are colored in red. Although, as with knots, a high number of visits was paid to the location of entry into the arena by a large number of mice, a high number of visits was also observed along the wall. While paying a high number of visits to knots, knot locations were not singled out by the number of visits paid to them (they were singled out only by knot-scribbling). The last session of the 5th mouse was not recorded because of a technical failure
(4.52 MB TIF)
Contour plots representing cumulative dwell time per location in all saline-injected mice. Each horizontal line represents, from left to right, all the sessions of a mouse. The colors represent dwell time. Low, medium and high dwell time locations are colored as in
(3.84 MB TIF)
Illustration of a knot established by a selected intact C57BL/6 mouse in a free setup allowing free passage between the mouse's home cage and a 2.5m diameter arena. A. Path traveled by the mouse within 1h during free exploration. B. Contour plot of path curvatures. Window size and colors are as in
(2.44 MB TIF)
We thank Noldus Information Technology for the use of their EthoVision system. We thank Prof. Yoav Benjamini for his help with the statistical analysis of the data and Dawn Graham for help in running the experiment.