ON-LINE CyberRat PURCHASE |
ORDER EVALUATION COPIES |
USER'S MANUAL |
PRODUCT INFO | COMPANY INFO | TECH SUPPORT | CyberRat RESEARCH | FAQ's
Ray and Brown (1975) were among the first to investigate behavioral patterning and organization in rats habituating to their first exposure to an operant chamber. In fact, this publication pretty much establishes the Behavioral Systems research foundations that was eventually used to create CyberRat. Unfortunately, the technology available in the early 1970's required Ray and Brown to use behavioral sampling techniques which only approximated behavioral sequences and associated unconditional/conditional behavioral probabilities. Durations were also not possible, since occurrence within the "five-second observation / 15 second recording" window was the only parameter recorded. Thus to make the new version of CyberRat as realistic as possible, a new series of investigations was conducted on behavioral system dynamics during first exposures of rats to being placed in an operant conditioning chamber. This allows us both to establish parametric ranges applicable to the simulation and to illustrate to interested CyberRat users exactly how behavioral systems principles are being modeled via the combination of digital video and controlling computer algorithms. In other words, how CyberRat was created and how it works.
Two animals were videotaped for 30 minutes, and a third subject for 45 minutes during their first exposure (habituation) to being placed in an operant conditioning chamber. All animal were approximately 3-4 months of age. Behavioral observations were made using a relatively mid-grained behavioral category system which included:
Initial recording were made with a digital video camcorder. These digital tapes were subsequently transfered to computer and edited into successive 15 minute compressed digital video recordings saved on recordable CDs. Compressions were set for fifteen-frame per seconds. These digital CD recordings of each subject were made with time markings appearing graphically depicting date, hour, minute, and seconds. An in-house authored software system was used to synchronize computer video play times with this graphical clock, thus allowing for a relatively accurate interpolation of actual frame numbers as well. This software system was then used to apply the above coding system to the synchronized digital recording, resulting in a file which identified each successive state of behavior and the time it was initiated. Subsequent analyses of these data included a sequential (kinematic) matrix reflecting the number of each preceding-succeeding behavioral sequence. These numbers were then used to generate both unconditional (independent of preceding behavioral state) and conditional (specific to each preceding behavioral state) probabilities for all above behaviors.
Finally, a Cohen's Kappa (Bakeman and Gottman, 1989) was calculated periodically to assure ongoing reliability between two coders.
Both dynamic and static analyses were accomplished for each subject and for the combined records of all subjects. Static analysis combined all data within the entire session into one summary of the total session. Dynamic analyses included measurement of various behavioral parameters within each successive 5 minute window of the 30 (and 45) minute sessions. We will first discuss static and dynamic summaries for all subjects, across all sessions, to form a general description of behavioral dynamics prior to discussing detailed individual change dynamics across sessions for each subject.
Global Behavioral Organization during Habituation: Kinematics
Kinematic flow charts illustrating conditional probabilities of behavioral sequences based on the above category system were determined for all subjects combined. The Kinematic Flow Fhart below describes only the highest probability paths of organization by dropping out all conditional probabilities with values less than .05 (a complete chart for all behaviors and all probabilities is available by clicking on the kinematic diagram below to enlarge its details).
The most frequently occurring behavior illustrated in this Kinematic Flow Chart is Exploring, which accounts for 40% of all initiated behaviors for the entire 30 minute session. From Exploring behaviors, subjects engage in Move 43% of the time, followed by Rearing and Dipper Entry each at approximately 10%. From Move and Rear, the animals are most likely to return to Exploring, with a probability of Move-to-Explore being .94 and Rear-to-Explore being approximately .40. Approximately 90% of all sequences involve these three behaviors.
Global Behavioral Organization during Habituation: Unconditional Probability Dynamics
As a means for better depicting unconditional behavioral probability dynamics and how they change across the duration of the session, we collapsed many associated categories into an even more "macro-level" description. This was accomplished by grouping our coding categories described above into "macro" behavioral categories of related "behavioral families" as follows:
Inactive Behavior (rest and freeze)
Object-Oriented Behaviors (bar press, bar touch, dipper entry, and object touch)
Self-Oriented Behaviors (groom self, bite self, scratch self)
Spatially-Oriented Behaviors (move, explore, and rear)
Subsequent figures illustrate unconditional probabilities for each of these "macro" categories across each successive 5 minute window of the 30 minute habituation session for all subjects combined plus a subsequent 15 minutes for subject A4 only.
These unconditional probability graphs illustrate that subtle changes occur in behavioral probability across the duration of the habituation session. They reveal relatively high probabilities of object-directed behavior at the start of the habituation session up to minute 15. At the same time spatially directed behavior is maintained at a high probability, with a minor dip and rebound at the end. With self-directed behavior we see a gradual increase until minutes 16-20, and afterwards it declines. Inactive behavior, while brief, does increase in probability towards the latter half of the habituation session.
Subject A4 data have been graphed with 15 additional minutes to illustrate trends after the collective data end. Self-Directed behavior continues the group trend seen in the previous graphs. A resurgence of spatially directed behaviors as well as in object-orientated behaviors appears in these last 15 minutes.
From these graphs we can thus describe the behavior of an average subject when placed in an operant chamber over a 30-minute period. The subject engages initially in many spatial explorations, but object-oriented behaviors are also relatively high in probability early in the period. This object-orientation gradually declines and is first replaced by self-directed behaviors and eventually is replaced by an increase in spatially directed behaviors.
Global Behavioral Organization during Habituation: Behavioral Flow (Velocity) Dynamics
Ray and Brown (1975) first introduced the idea of measuring the rate at which behavior changes, regardless of what behaviors are involved. Over the years this measure has been referred to as either "behavioral flow rate" (Ray & Brown, 1975) or "behavioral velocity" (Ray & Delprato, 1989). In essence, it is a general inverse reflection of the durational aspects of behavior. When behavioral velocity is high, each behavior is relatively short in duration. When behavioral velocity is low, behaviors tend to be of a longer duration. Selective behaviors can impact this general measure, of course. For example, grooming tends to be much longer in duration than movements from place to place as well as rearing in upright exploration. As these behaviors change in probability with respect to one another, one also might expect general behavioral velocity to reflect that. Of course specific behaviors can, and do, also change in relative duration within a category (Ray, Upson, and Henderson, 1985). The average behavioral velocity for all subjects across each successive 5 minute window is depicted in the graph below.
Behavioral velocity measures depicted in this graph reveal a gradual decline in the rate of change from behavior to behavior. Stated another way, behavioral durations are gradually increasing across the 30 minute session until the last 5 minute period, at which time velocity increases again. As noted, this is quite possibly due to the fact that the animals increase spatial exploration during this last period, and such behaviors are typically of shorter duration than self-directed behaviors such as grooming.
This Study and CyberRat V 2.0
The coding system used in this study was also defined as a general "index" so that it might be used as a general "search" engine for very fine-grained behavioral codes used to edit a new and improved digital video catalog for CyberRat V 2.0. In the actual working version of CyberRat, directional orientations and place within the chamber have to be taken into account, and many "behaviors" at the mid-grain level are broken into components suitable for micro-definitions and their associated implications for video "splicings" into alternative "branches" of behavioral variations that may then be used to more economically recreate macro categories. Thus a tightly executed "clockwise circle" might actually be made up of an appropriate sequence of "quarter-turns." Initial "quarter-turns" might alternatively branch into "go to dipper" behaviors during a shaping process, wherein reinforcement initially "interrupts" the more complete circle that is eventually being taught. An astute reader will have already caught on to the fact that "turns" aren't even a categorical element of the coding system being reported presently, much less "clockwise" or "counter-clockwise" turns. In the above system, these would all fall under the "move" category.
This ability to "translate" from familial molecular categories by collapsing them into "parent" macro categories is, of course, one of the explicit properties of higharchically defined subsystems within systems theory. As such, we are also able to deliver within CyberRat the ability to apply to its video archive, for any session, a similar coding scheme as that being reported here. This is the "mechanic" for the new feature that graphs what is happening to relatively unconditional probabilities of such behaviors during various "moving windows" of a session.
Copyright 2001-2002
(AI)2, Inc.
all rights reserved