For the Defense - Vol. 7, Issue 4 - 7

the death a homicide (35%) than an accident (6%), even
though her history and injuries were identical.6
The Backlash: Hostility and False Accusations
While alarming, these results were unsurprising given
decades of research on cognitive bias in human decisionmaking,
including in forensic science and medical diagnosis.
Why should post-mortem decisions be any different? But
upon its publication, the response from medical examiners,
including NAME leadership, was swift and severe. They
accused us of data fraud, ethics violations, and tampering
with the peer review process-claims that were thoroughly
investigated and found to have no merit. Even so, NAME
demanded that the paper be retracted and threatened legal
action.
All the while, the coincidental timing of its publication-
during the trial of Derek Chauvin for the murder of George
Floyd-only fanned the flames. Our study received ample
media attention because it cast doubt on medicolegal
opinions that were central to the case-and indeed, some
of our most vocal critics were concurrently serving as expert
witnesses for Mr. Chauvin's defense.
To quell the controversy, the journal offered to publish
commentaries on our paper from any interested parties, of
which they ultimately received 11, ranging from laudatory
to vitriolic. As I explain below, these commentaries also
shed light on some medical examiners' misconceptions and
decision-making processes, which proved arguably more
damning than the study itself.
The Commentaries: Debunking Myths about Bias
The first letter from Dr. James Gill, then-President of NAME,
argued that our data were invalid because some respondents
may not have been " board-certified forensic pathologists. " 7
Implicit in this argument is the widespread myth that training
and experience are prophylactics against cognitive bias, which
is unfortunately not the case. In fact, some researchers have
argued that expertise actually begets biasability via greater
reliance on heuristics and other mental shortcuts.8
This argument also exemplifies a phenomenon called the
bias blind spot, wherein people more readily detect bias in
others than in themselves, or in this case, believe that bias may
affect some pathologists, but not others. Similarly, in a 2017
survey, we found that 71% of forensic scientists saw cognitive
bias as a problem for the forensic sciences as a whole, but only
52% saw it as a problem in their own discipline, and only 26%
felt that it could affect them personally.9
Other letters accused us of seeking to portray respondents
(and by proxy, their colleagues) as racists, but as noted
above, cognitive bias is not the same as intentional prejudice.
Moreover, because we varied both the child's race and their
caretaker between the two versions of the vignette, we
cannot be sure if respondents' decisions were affected by race,
caretaker, or both-but in any case, it is clear that they were
unduly influenced by extraneous non-medical information.
Relatedly, some critics argued that the caretaker's identity is
relevant to determining manner of death because population
data show that genetically unrelated men more often abuse
children than do female family members. There are two
problems with this claim. First, these statistics may themselves
be a product of bias, i.e., if unrelated male caretakers are
more often identified as abusers because they are more often
suspected of abuse, similar to how discriminatory " stop-andfrisk "
practices created a self-perpetuating bias against Black
individuals.
Further, this argument suffers from the ecological fallacy,
namely that aggregate data cannot be used to make
inferences about individual cases. For example, men are, on
average, taller than women, but simply knowing a person's
height does not allow to you infer their sex. Similarly, most
Vol. 7, Issue 4 l For The Defense 7
For the Defense - Vol. 7, Issue 4

Table of Contents for the Digital Edition of For the Defense - Vol. 7, Issue 4
