Ann Burgess has spent more than five decades studying the people who commit the most violent crimes and the people they hurt. She helped create the scientific foundation for behavioral profiling, co-developed the concept of rape trauma syndrome, and worked with the FBI’s Behavioral Science Unit on the first systematic study of serial killers. At ISHI 36, she delivered a keynote examining where that research goes next, and what it will take to move from understanding violence to preventing it.
Her co-presenter was her granddaughter, Alexandra Burgess, whose work in data science and algorithm development is bringing new computational tools to the behavioral questions Burgess has spent her career asking.
Where It Started: Victims First
Burgess’s path into criminal research began not with offenders but with victims. As a faculty member at Boston College, she partnered with sociologist Lynda Holmstrom on a study of rape survivors, work that required approaching hospitals in Boston directly, since victims rarely identified themselves. Boston City Hospital was the only institution that responded.
What emerged from that research was the concept of rape trauma syndrome, a framework for understanding the psychological aftermath of sexual assault. The study put Burgess on the map as a leading voice on victimology, and it eventually drew the attention of the FBI.
The FBI Study and the Origins of Behavioral Profiling
When the FBI began developing what would become behavioral profiling, they needed outside expertise. William Webster, then-director of the FBI, directed the Academy to begin teaching rape victimology, and that mandate brought Burgess in. What she found when she arrived was a group of agents who were already doing something remarkable: sitting around a table with crime scene photographs, going around the room, and describing the characteristics of the person they believed had committed the offense. Not a name. A profile.
The agents who drove that early work included Bob Ressler and John Douglas, along with Howard Teten and Patrick Mullany. Burgess served as the methodologist on the research that would follow.
The study ultimately included 36 subjects. The sample started at 82 potential participants, was narrowed to 36, and the majority had killed three or more victims, the definition of serial killer used at the time. Agents would travel to prisons and conduct interviews using an 87-question protocol. Every answer was cross-referenced against police reports and official files, because, as Burgess put it, the subjects didn’t always tell the truth.
“You learn they don’t tell the truth all the time. So we wanted to match what they would say about the crime, and then we would have the police report. So we could kind of cross match it.”
The protocol change that made two-agent interviews mandatory came directly from experience. During one session, an agent found himself alone in a room with Ed Kemper after the signal button went unanswered. Kemper told him he could twist his head off in a minute and be the most powerful person in the prison. The agent shifted from interviewer to hostage negotiator. After that, no agent went in alone.
The Pathway to Violence and the Prevention Question
The original goal of behavioral profiling was to read a crime scene and generate suspect characteristics. The goal Burgess is working toward now is earlier: not who did it, but whether it can be stopped before it happens.
She describes a six-step pathway to violence. Every person who commits a targeted act of violence has a grievance. For most people, it stays there. For some, the grievance becomes an obsession, then a plan, then a target. Burgess argues that the third step, when the obsession becomes active planning, is the intervention window.
Social media posts and manifestos are central to that effort. Offenders write things down. They post. They document their thinking in ways that leave a trail, if researchers know where to look and how to interpret what they find. The Red Lake shooting, which occurred in the early 2000s as the internet was just beginning to scale, is one case Burgess’s current research is examining: what do the posts look like then, versus 2010, versus now? What changes in the language as the planning advances?
“We’ve moved from serial killers to mass shooters.”
Three Generations, One Research Program
The keynote Burgess delivered at ISHI 36 was a family project. Her granddaughter, Alexandra brings data science and algorithm development to questions Burgess has been asking for decades. Her daughter, a pediatric nurse and academic, contributes expertise in social media and adolescent behavior, including research on students answering phones in their sleep.
Burgess is direct about what each generation brings that the others don’t.
“They all need an Alex, you know, to help them move the data onward.”
The jail diversion project Alexandra is currently leading illustrates the approach: 15 years of 911 call data, analyzed to determine when to dispatch a police officer versus a mental health professional. The algorithm is generating answers. The answers still require human interpretation.
“You still, for this kind of research, you have to have human interpretation. It’s just one more tool to use.”
AI in the Courtroom: The Menendez Case
Burgess testified for the defense in the first Menendez trial — the one that produced a hung jury, with half the jurors finding merit in the defense’s argument that something was deeply wrong in the family. She has 16 childhood drawings made by Eric Menendez that have never been published. The lead defense attorney, Leslie Abramson, chose not to use them at trial, judging that a jury would dismiss drawings made by a juvenile.
Burgess has since run those drawings through an AI analysis alongside an art therapist, with herself providing a psychiatric interpretation. The result was agreement between the human interpreters and the machine on certain findings. What to do with that agreement, whether it strengthens or complicates testimony, is a question the field has not yet resolved.
“You bring in everybody to testify and you tell the jury what you found, and they have to make the decision.”
What Forensic Scientists Should Know About Communicating Complex Evidence
Burgess has testified as an expert witness across decades of high-profile cases. Her advice on courtroom communication is specific: answer the question as simply as possible, do not volunteer, and know that one small error can compromise everything else you’ve said.
She points to the Andrea Yates case as an example of how a credible expert can undermine their own testimony by stating something that can be checked and disproved. The lesson applies directly to forensic scientists presenting DNA evidence, algorithmic findings, or any evidence type that requires a jury to trust both the methodology and the person explaining it.
“If you’re going to go into court, you have to be able to explain it simply… Communication is critical.”
Her final point is about bias: every expert has it. The question is whether you know what yours is and can account for it.
“You have to give an opinion and the basis for my opinion — it’s got to be based on science.”
For the forensic DNA community specifically, Burgess sees a clear parallel between what DNA has navigated over 30 years in the courtroom — protocol challenges, credibility questions, the long process of building jury trust — and what behavioral evidence and AI-assisted analysis will face. The lessons are transferable. The work of making them transfer is still ahead.