Computer vision tracking of animal activity has dramatically improved efficiency in behavioral neuroscience research, but many complex behaviors still require manual scoring by a human observer. In some cases, automated activity tracking provides most of the important variables but the researcher wants to supplement those data with manual observation. For example, a researcher studying anxiety-like behaviors in the elevated plus maze may use tracked activity data for generating basic variables like open arm time and then manually observe more complex markers of anxiety, such as stereotypic behaviors, fecal boli, and self-grooming.

In other cases, the experimental focus is a set of complex behaviors not easily classified by computer vision systems. For example, researchers studying behaviors like aggression, social interactions, or maternal care may find manual observation to be the most reliable form of analysis. In the future, machine learning will likely enable automated scoring of some of these complex behaviors, but for now manual observation remains a key method for behavioral neuroscience.

In the coming weeks, BehaviorCloud is introducing a simple, user-friendly interface for manual behavior observation and scoring. We provide all the basic tools you’re accustomed to using for manual observation, but we’ve made substantial improvements to the experience.

1. Observe and score video through your web-based account, so you can work from anywhere, anytime - even from a mobile device.

2. Set permissions to allow collaborators access to your experiment - multiple observers can work on scoring the same videos in parallel so you get reliable results in less time.

3. Securely store your videos in your BehaviorCloud account, keeping everything accessible and organized.

<img src="/images/blogimages/manual_scoring2-83d27817.png" width="100%" "Manual scoring configuration">