In the Mason group direct live observation has been successful (Bechard et al. 2010; Tilly et al. in press). Live observation has three main advantages; first, in terms of recording, it eliminates the time and economic cost involved in buying cameras and recorders, setting up equipment, and changing DVDs/downloading video files; second, the viewing angle can be altered in a flexible and ad hoc way to view any otherwise out-of-sight mice; third, when it comes to data analysis, figures are available for processing rather rapidly, instead of waiting for the extra phase of DVD-watching. Its disadvantages are potential observer effects (a habituation period of at least two days to acclimate mice to the observer; and also being quiet and slow-moving throughout the whole period of data collection is recommended); that fast, hard-to-see behaviors can be missed or mis-scored, because they cannot be played back slowly for analysis; and that extra steps need taking to ensure blinding (e.g. obscuring cage labels from the observer). Overall, however, for quantifying simple aspects of behavior and/or for feasibly assessing large numbers of cages at once efficiently, direct observation will often be more efficient than video.

 

Both regimes below worked well: the data were very internally reliable (as assessed by correlating mean values for odd days with mean values for even days, to check for consistency) and also had good inter-observer reliability (as asses between two blind, non-consulting observers recoding the same mice at the same time).

 

 

 

The observation period spanned the initial 2 hours of the active (dark) period. Families were directly observed daily, under dim red light, from PND 16 until PND 35. Each family was observed once every 10 minutes, for a total of 12 observations per day, and final scores were calculated as a daily percentage of observations. Behaviors were identified using a mixture of scan and focal sampling: a family was scanned initially to identify the focal mouse of interest (e.g. dam), followed by instantaneous observation of that individual, its behavior being scored using the ethogram in Table 1.

 

 

 

Adult mice were watched for 4 hours a day starting at 2 hours after lights-off. This was repeated for 8 days in one phase of the experiment and 12 days in another; in both periods data from odd and even days correlated very strongly showing that collecting data for half this number of days (thus 4-6 days) would have given equivalent information.

 

Cages were scanned in a consistent order by a quiet observer once every 10 minutes for 4 hours. As a row was scanned, instantaneous sampling was used to record patterns that were identifiable immediately and apparently normal ( e.g. sleeping, standing, lying alert). However, this was unsuitable for putative stereotypic behaviors which are identified by their degree of repetition. In such instances the observer switched to focal observations, activity being continuously observed for up to 15 seconds or until a regular pattern emerged (whichever was earlier). Behavior was classified into one of four broad categories (see Table 2).