A review (Pinizzotto, 1984) of criminal profiling methodology outlines some of the basic principles of profiling, and which cases are best suited for profiling.
I enjoyed this review as it utilised an actual case to demonstrate the process. Let's summarise the article before moving on to cautionary tales of the difference between deductive reasoning and inferential reasoning.
Pinizzotto points out that before any profiling can commence, the forensic psychologist allows evidence to inform the procedure.
A list of criteria for profiling to continue (Ault and Reese, 1980 as cited in Pinizzotto, 1984) contains some of the following:
Photographs of the crime scene
Details of the neighbourhood, ethnic and complex surroundings
Medical examiner's report
Background of the victim (including personality, lifestyle, ethnicity, scholastic achievements etc)
These considerations allow as Pinizzotto points out, allows forensic psychologists to answer the 'what' of the crime. To answer the 'why' is where a profiler enters the picture. If these two questions can be answered, then profiling will be able to answer the 'who' question, at least that is the plan of the profiler (Hazelwood, 1983 as cited in Pinizzotto 1984).
Deduction vs Induction
One of the biggest misconceptions of the Sherlock Holmes stories as with most detective fiction features a protagonist who has a particular skill with deductive reasoning. However, in truth, the conclusions are almost always arrived at via inference rather than the absolute and logical certainty offered by deduction.
A brief example will facilitate an understanding of the important difference.
Deduction relies on a conclusion being absolutely true if and only if the premises are true also. You cannot have a false conclusion if the premises are true.
To illustrate we'll use the classic argument:
All men are mortal
Socrates is a man
Therefore, Socrates is mortal
Here no new information is arrived at, and the conclusion merely reveals a fact hidden within the premises.
Induction is different, it seeks to add information that goes beyond that data we have. It uses prior experience to infer how future events will play out. In this case, the inference is perfectly acceptable but the ending, unlike deductive reasoning is never a given.
Here's an example from Bertrand Russell: A chicken is fed each day by a farmer, it may assume on any given day with the farmer's approach that the day is no different from any other - except for the day in which the chicken is killed.
Leaving aside Hume for a moment and the question of the validity of inductive reasoning, we all use induction every day and with great success. However we may be wrong, our true premises do not guarantee the truth of the conclusion.
Detectives in literature often have a list of characteristics that they append to a suspect, with high probability, and yet there is in fact no reason to believe that their conclusion is the right one.
If the great Sherlock Holmes sees tobacco ash matching some brand, and we allow that the killer smoked this brand, we cannot assume that the killer will be caught smoking that same brand - for he may have only smoked these due to access or availability. In this way, our suspect could not be given the liberty of a 'not guilty' verdict because of one eliminating factor and this apparent discrepancy.
What inductive reasoning can teach us about profiling
In this way, a profiler may skilfully apply their trade to the unknown perpetrator, but however subtle and ingenious the inferences - they are not guaranteed to reach the appropriate conclusion. Thus profiling becomes a game of probability, which is absolutely fine as long as we don't mistake the game being played for a deductive one. How do we ensure that profiling is kept clean objectively speaking, we use the tools already reserved for scientific inquiry? Not verification for we will verify anything we want to believe in. Instead, we rely on Karl Popper and falsification.
With these warnings in mind, we can enjoy and marvel at the work of profiling as an astute method for drawing reliable inferential conclusions, whilst all the time standing ready to alter our conclusions if needed.