What is behavior? Most simply, it is just movements of an animal (although sometimes the lack of movement can be as significant as an intricate courtship ritual). As such we need to be able to record accurately two things about an animal or a part of its body: where it is and how its position changes through time. The ways of doing this description vary with the technology available. When all we had was our ears and eyes to observe, and pen and paper to record, the movements tended to be summarized into ‘patterns’ of behavior, considered more or less fixed in sequence and expression, so that an observer could simply write down a pattern name (e.g., ‘courtship’) and use it as a shorthand for a whole series of intricate movements. Although this shortcut works well for some animals with very stereotyped behaviors that are fairly conspicuous and not too frequent, it is not appropriate for animals with flexible behavior, for very brief behaviors that are likely to be missed, and for behaviors that are too frequent to allow time for recording. For instance, the songs of birds tend to be fairly constant in form within an individual, so it is fairly easy to just record the number of times per minute that a male bird sings on his territory. In contrast, monkey vocalizations are ‘graded’, with many intermediates between extreme ‘pure’ types, so that judging what a given call represents is difficult by ear.
 
 

The introduction of tape recorders facilitated both the recording of frequent or brief behaviors (no need to write them down), increased the accuracy of measurement (could play a tape over and over to measure a given interval), and allowed more complex analysis of the form of vocal behaviors (you will learn more about this in a few weeks). With the widespread introduction of video recorders, we can now easily analyze image (that is, spatial) information with a degree of precision not available even a decade ago. This facility opens up new realms of questions that are difficult or impossible to address using the unaided eye: simultaneous spatial relationships among many individuals in a social group, subtle differences in behaviors, recording of very brief behaviors, etc. Perhaps the best recent example I know of is Marc Hauser’s demonstration that the right side of a monkey’s mouth opens a few milliseconds before the left side when its starts to vocalize, thus demonstrating brain laterality for communication in non-human primates, as there is in human speech.
 
 

Regardless of the type of recording method used, you will be faced with several decision about how to take your data: choice of individual(s) to sample, how often or long to sample, and how to break up a continuous stream of activity into discrete behaviors.
 
 

Choice of individuals to sample. If you are interested in what one individual is doing, you would use a 'focal animal' sampling technique, in which you record data from only one subject. If instead you are interested in what the average individual in a group is doing, you would use a 'scan' sampling technique, in which you record the behaviors of all visible individuals as you scan across the group. A serious problem in choosing a subject or subjects is observer bias. It is very easy to be more attentive to animals that are moving than immobile, or ones that are doing more interesting behaviors versus those that are doing less interesting behaviors, etc. If you choose your subjects non-randomly, then your behavioral averages will reflect your biases as much or more than the behaviors of the animals themselves. One method to get around this bias is to have a systematic design for choosing the subjects, such as listing the individuals (if they are known as individuals) in a random order and then rotating down the list in sequence.
 
 

How often or long to record? Even if the subject is randomly chosen, this does not eliminate observer bias in recording more interesting behaviors, so it is often a good idea to develop a systematic way to record WHEN to note behaviors. One solution is to take data (or start taking data) ONLY at predetermined time points (e.g., exactly at 5-minute intervals). Thus, you are not tempted to begin a sample only when the animal is doing something ‘interesting’.
 
 

The major consideration in sample frequency or duration is the temporal pattern of the beavior itself. If the animal does the same thing for relatively long periods (say, one minute or more), then you can record each behavior in a CONTINUOUS sample, noting only the time that one behavior gives way to the next (called a TRANSITION) along with the type of behavior as it changes. If behaviors change more rapidly, then it may impractical to record all transitions, unless you have a tape- or video-recorder. One alternative is TIME-INTERVAL or INSTANTANEOUS sampling of behavior, in which you note down what the animal does only at predetermined time points. For instance, you might decide that you can only write down something once a minute. In this case, after you find an animal, record what the animal is doing each time the second hand on your watch exactly reaches the minute mark (a repeat count-down timer, available on some sports watches, is very useful for this type of sampling). If you can record more rapidly than once per minute, then you can reduce the interval between observations. This kind of sampling is easy to perform, but is often incorrectly analyzed statistically (each observation is usually not independent of the next, so standard statistical tests are not valid).
 
 

The previous systematic sampling schemes work well for behaviors that occupy more than a few percent of the animal's total time. However, they are not suited for recording rare behaviors, especially if these are of short duration, because they will rarely occur 'on the mark'. In this case, you have no choice but to record the behavior of interest every time you notice it, even though there may be times that you missed it because you were not paying attention. Such opportunistic or AD-LIBITIUM sampling, while less methodical than the other methods, works well for conspicuous behaviors (fights, alarm calls) which you are unlikely to overlook. Finally, if you can't see the animal for long periods, the behavior you wish to watch is of very short duration, and it is inconspicuous to boot, give up and work on something else!
 
 

What to record? There are two difficulties in obtaining an accurate and unbiased description of the behavior of any animal. First, you have to be observant. This is definitely an acquired skill for most people. Being observant means more than just looking; it means always asking questions as you look, so that you notice rare or odd occurrences that others merely see. The more you watch behavior of a given species, the better you get at it. Once certain patterns emerge and repeat themselves, deviations from these patterns will be more obvious.
 
 

Second, you must decide how to describe what you see. What words will you use to name a given type of behavior? How will you decide what movements of the animal constitute a distinct, nameable behavior pattern? These are not questions with obvious answers. Indeed, part of what make behavioral studies difficult to repeat is that different observers do use different definitions for a given named behavior pattern, or include different animal movements under the same name. It is often tempting to name behaviors according to their apparent function, but this practice is risky, as what you perceive as the function may differ markedly from the actual immediate stimulus or long-term goal of a behavior. For instance, finches foraging on the ground frequently cock their heads, aiming one eye skyward. This behavior could be called 'predator-vigilance', as it seems to be well-suited to discover birds of prey which might attack the finch. However, study of this behavior shows that it can occur also when looking for other finches (if the individual is afraid of getting chased away), and looking for new foraging places. Thus, the behavior is best labeled by a name that merely describes the movement (such as 'look up') and avoids any functional interpretation, which can always be added later on when the pattern of occurrence of the behavior is analyzed.
 

Your ‘field study’ for this lab exercise will consist of taking data on monkeys and coatimundis ( a relative of the racoon) in Iguazu National Park, Argentina. Unfortunately, SUSB would not cover the airfares, and we could not get permission to import the animals to our laboratory, so instead you will work with video footage taken of these animals by a variety of photographers, both professional and not (many thanks especially to Mario Di Bitetti for use of his videotapes). I have selected as much good video footage as will fit on one CD in compressed MPEG format. The use of computer files, of course, allows us to analyze the resulting information digitally.

There are a variety of general categories of behavior shown in different files: foraging, feeding, resting, courtship, grooming, rapid movement through trees. Because I selected these based on video quality and availability, they cannot be used to construct a general picture of the ‘average’ behavior of either species (why not? How would you design a sampling method to obtain such an overall average). Thus, we have two alternatives: 1) we can divide students into groups of behaviors and allow each group to decide what is interesting to analyze about each behavior type, or 2) we can decide on some behavioral question which can be analyzed in all of the general behavior types – possible candidates for the latter include the use of the prehensile tail by the monkeys, the frequency and duration of ‘scanning’ during different activities (see above for finches), and spatial relationships between individuals during different activities. For some of these questions, the most important variable will be time intervals, for others spatial data will be necessary. We will demonstrate the use of two different computer programs that allow you to measure time and space fairly precisely on the video ‘screen’. Unfortunately, no one program does both space and time measurements well.
 
 

We have enough CD copies to allow every pair of students to work together. After an orientation in the computer teaching lab, we will discuss possible questions and let you get to work becoming familiar with the programs and taking data. After a couple of hours, we will stop taking data and return to the course lab room to discuss how to analyze the data. If time permits, one member of each pair will present a few sentences to the class about what they found out.