Like a fingerprint, no two faces are exactly alike. Recognizing and identifying key people related to a crime (including the victim, perpetrator, and other persons of interest) are often the first criteria that must be met to advance a forensic investigation. Investigators and forensic scientists draw on a number of branches of science that study faces, such as anthropology, psychology and art to develop useful tools to find these key people.
In Jackie English’s case, one of the tools at the squad’s disposal was a sketch that depicted a man seen speaking to Jackie several days before her disappearance. The sketch was based on a description provided by Ms. Harrison, who later became a victim of harassment and assault due to her involvement in the case. Because she was attacked after providing information to police, there was speculation that the man who attacked her was the same person who killed Jackie. When the TCAK Squad identified two potential suspects in Jackie’s case, we needed to systematically and scientifically assess whether the sketch pointed to one of these people more than the other. To do that, I approached Dr. Cathy Mondloch of Brock University, who is a specialist in the psychology of facial perception.
We see faces all the time, such as the faces we know and recognize such as family and friends, to the hundreds of strangers that we pass every day as we go about our daily lives. Faces convey a wealth of information in our daily social interactions – information about the identity of individuals, their emotions, and their direction of gaze. When creating a composite sketch, the most crucial component of an accurate facial composite is the quality of an eye witness’ memory, and their ability to describe what they remember to a competent artist. Describing the face of someone who you may only have seen in passing is an imperfect process. Human memory can play tricks, erasing certain details and amplifying others, and the events that co-occur at the time of the sighting can heavily influence your perceptions.
Our analysis of the sketch was based on the premise that we only notice and remember facial features that stand out as “unusual” from an average face. When thousands of photographs of individuals (controlled for sex and ancestry) are superimposed and amalgamated into a single face, you get what is described as an “average face”. These artificially created average faces help psychologists study how we perceive faces, including how and when we notice individual features. Rather than notice every single feature of a person’s face, your brain only picks up the features that are different from average in a meaningful way. It makes sense then, that a composite sketch is based on features that stood out to the witness as unique, or significantly different from average.
To measure just how “different” the sketch from the English case was from an average face, markers were overlaid onto the sketch that correspond with standard facial landmarks. These landmarks, such as the corners of the mouth, outer and inner corners of the eyes, top of forehead, and center of lips are reference points, and the distance between these established landmarks can be used to quantify facial proportions and define “uniqueness”. To qualify as unique, a feature must be statistically different than the average, defined in this case by two standard deviations larger or smaller than the average for that feature. This definition of statistical significance acts as a proxy for the degree of uniqueness that would stick in the witness’ memory as a way to describe the person of interest.
All of the features measured on the sketch were found to be within statistically normal proportion to each other. This means that any features that seem abnormal based on the sketch are unlikely to be the result of incompetence or inexperience on the part of the artist, and reflect natural variation in facial features in the individual depicted. The sketch was found to differ from normal to a statistically significant degree in seven ways.
When the facial features of the squad’s two potential suspects were measured, neither differed from normal in exactly the same pattern as the person depicted in the sketch. There was, however, a greater degree of similarity between our main person of interest and the sketch than the second person of interest and the sketch: When we compared the first person of interest, of seven characteristics that could be compared, four differed from normal in the same way as the sketch and no features contradicted each other. The second person of interest was much less likely to be the person in the sketch, as his tall forehead height contradicted the short forehead seen in the sketch. Statistically speaking, this meant that there were at least four standard deviations between the forehead height of the sketch and of the second person of interest.
Even with the compelling concordance between our first person of interest and the sketch, we can’t say for sure that he is the person the sketch was based on, or that the person in the sketch is Jackie’s killer. Despite this, we believe our analysis represents the first time research from facial perception has been used to scientifically analyze a composite sketch, and compare a person of interest to a sketch in the absence of the witness who gave the initial description.
To see how a civilian squad tackles unsolved cold cases, and learn about new innovations in investigative methods, watch the new True-Crime docu-drama series To Catch A Killer on the OWN Network.
Read some of the press coverage of the episode: