New Techniques 
Home Publications Facilities New Techniques Photo Album BabbleLog Bird Research Baby Research People


 Described below are some new ways we are going about studying social behavior.

Voice Recognition

We now collect observational measures of singing and social organization using IBM ViaVoice speech recognition technology. Spoken information is sent by wireless mic to the lab, received by ViaVoice, transcribed into text, and downloaded into a database. The database checks for errors, organizes and summarizes the data all automatically. It has proven to be an incredibly accurate system that provides us with a hands-free interface and removes manual data recording and data entry time. For more information on the system, click here. And for even more information, click here. This capability has allowed us to vastly increase the quality and quantity of behavioral data we collect and consequently has created the ability  to investigate the development of social/communicative behavior in realistic social settings. In our first large flock study over an 8 month period four observers collected approximately 32,000 data points and entered the data  in a data base. Today four observers can collect and analyze a comparable data set in about 20 days. The voice recognition data is all time stamped so that we can identify patterns of action and reaction that are related in time. The continuous flow of time stamped data led us to develop a new statistical tool to identify recurring sequences of behavior not easily seen with traditional statistics (see below).              

Modified Triadic Census

Studies of learning or the development of behavior typically measure the behavior of one or two individuals at a time but learning in nature commonly occurs when multiple individuals are present and interacting with each other as the eavesdropping literature shows..  To address this real world problem we use a modified triadic census that parses behavioral data from a flock into interactions among three individuals at a time. The figure on the right shows all possible interactions between three individuals and is taken from Wasserman and Faust (1994). Unlike previous implementation of the triadic census we use the occurrence of a particular behavior as an initiating event for the formation of a triad and require that subsequent events occur in immediate temporal proximity to be considered part of a triad. The implementation of the triadic census produces a distribution of triad frequencies that is very sensitive to individual differences and variation in social structure. The figure below shows one such distribution created after male singing to females.                  

The figure on the left shows the triads created when the initial event was a male singing to a female. The X axis shows that only 11of the 16 triad types were present. The Y axis, on a log scale, shows the frequency of the triad types. It can be seen that the most frequent triad, the first bar on the far left, was when a male sang to a female and nothing happened. It can also be seen that the three next most frequent events are (from the left) triad types 3,5,7. In all of these triads after the initial song the male either sang to another bird, was sung to or interacted reciprocally with another individual. While not shown in this figure we can then break out the exact type of interactions within the triads. When we do this we see that  virtually all of the subsequent interactions in the 3,5 and 7 triads are with males. Thus,  in this particular flock, when a male sang to a female the most common event to follow was male-male singing.



This is Selfish Jeanne. She is a robotic female cowbird programmed to move her head, move back and forth along a perch and to wing stroke. She will be used to interact with live cowbirds. Robotics give us the power to have control over one aspect of a dynamic system.  Click here start 2.8MB movie of Jeanne


Computer Simulations            

Requires Macromedia Flash 6

Requires Macromedia Flash Player

In collaboration with Anne Smith, a former PhD student in our lab, we have begun designing agent-based simulations where individual simulated organisms are programmed with a series of behavioral rules and then allowed to interact with one another. The patterns in these interactions then can be documented. Such simulations will be used to test the rules of social interaction that cowbirds may be using that produce the social profiles we have documented in the aviaries. 

Two animations of the Virtual Observer simulation showing male and female organism assortment patterns under high  and low bias bias to moving toward same sex individuals (males are black, females are brown). Requires Quicktime 5 or 6


Hand Raising Baby Birds

Egg-to-egg development

Although we have hand reared cowbirds in the past, we required the assistance of a host species to feed the babies for the first five days. In the left frame a baby cowbird is fed by a canary.  This year the lab tried and succeeded in raising babies from the egg, straight from our incubator. The eggs were obtained from females in our aviaries. The females lay eggs in nests that we build. Pictured to the right above is one such nest. The speckled egg is a cowbird egg. The other two "eggs" are yogurt covered peanuts. Female cowbirds prefer to lay eggs in nests that contain eggs. The yogurt covered peanuts work very well as substitute eggs. Before laying her own egg the female will remove a host egg as pictured to the right.


 The aviaries are observed extensively during the breeding season so we have extensive social histories on each individual female and her mate. We have found that female fecundity and egg viability varies greatly across different social environments. We now have the ability, using hand rearing and DNA analyses, to extend individual and group social histories across generations: to see how the growth and behavior of the young correlate with the social competencies of their parents or their social group.


Pictured to the right  is an incubator tray with each row showing the egg output from 5 of our  aviaries spanning a 12 day period: all of the aviaries housed about the same number of females but egg production varied widely. After the eggs hatch, the babies are fed every 15 minutes from 6:00am to around 11:00 PM


Pictured below is the same baby cowbird photographed on days one, ten, fifteen, twenty one and forty eight.

Day 1


Day 10


Day 15 with leg bands OD (orange dark blue)



Day 21



Day 48 a female

  Day 79 for OD & day 72 for LY (light blue yellow) a male.  LY is attempting to sing to OD.

Recording Babies


We use small wireless microphones (FMR-150, Telex Communications) to obtain high quality recordings of infants' prelinguistic vocalizations. The microphone and transmitter are fitted in overalls that the infant wears throughout the play session.