There is a growing consensus in the forensic science community with respect to the existence of cognitive biases and the need for context management (e.g. Mattijssen et al. 2016; Dror et al. 2015). As mentioned in chapter 4, the most common solution proposed is the creation of a context controlled environment; in which the analyst is separated and blinded from potentially biasing information. This is important because the opinions of forensic scientists are considered impartial and unbiased, and not influenced by elements of the case that have no relevance to the scientific process
(Dror 2014a). The findings from Experimental 3 showed that ‘extraneous’ grave clothing associated with skeletal excavations had an impact upon initial beliefs, interpretations, the confidence level of judgments, and the subsequent assessments of skeletal remains.
Unexpectedly experiment 3 showed that early exposure of a strong indication of a certain outcome did in fact ‘bias’ the interpretation of non-ambiguous male skeletal casts. In contrast to previous research where cognitive and contextual effects have been shown to be more prone and make an impact when making decisions on ambiguous skeletal remains, or ‘difficult’ and challenging finger marks or DNA (e.g. Nakhaeizadeh et al. 2014a; Nakhaeizadeh, et al. 2014b; Dror et al. 2011; Dror et al. 2006), this study shows that contextual effects can also have an impact on non- ambiguous evidence. The results also highlight that not all contexts in a given situation appear to affect the final assessment made. For example, the findings only showed a bias cascading effect on subsequent assessment when exposed to a very strong context (female clothing on male skeletons). This shows (compared to experiment 1 and 2) that when dealing with non-ambiguous skeletal remains, only a very ‘strong’ context altered the interpretation made by the participants when assessing the male skeletal cast. Moreover, it highlights that the level of influence of contextual information will depend on:
• The nature of the task (Dror 2014a) and the ambiguity level of the given characteristic of evidence being interpreted (the difficulty of the judgment). • The strength of the context in which the decision is made, as well as on the
direction of the ‘bias’ (Dror 2016).
There have been some solutions proposed that seek to shield forensic experts from being ‘biased’ at an early stage of an investigation. Some have suggested a separation of different roles whereby the expert collecting the evidence not necessarily being the same one analysing it in the laboratory (e.g. Kassin et al. 2013; Saks et al. 2003; Krane et al. 2008; Dror 2014b; Dror et al. 2015). In forensic anthropology it is important that the forensic anthropologists are present on site in order to help preserve, excavate, and document the skeleton in situ, and mitigate the potential for the loss of important information pertinent to the anthropological assessment of the remains. Whilst it is recognised that it is important to utilise a combination of
different types of evidence in the creation of a biological profile, this carries the risk of the anthropologist being exposed to ‘extraneous information’ at a very early stage of a forensic investigation. This needs to be considered and measured when developing approaches for scene management, evidence collection, and assessment. It is also important to consider that grave artifacts and items of clothing associated with skeletal burials are evidence in their own right. The results from experimental study 3 showed that clothing was influential. The results therefore indicates that it may be beneficial for these items to be considered separately from the assessment of the bone features to reduce the potential for cascading bias, which may arise as a result of irrelevant information cascading from one stage to another (Dror et al 2017).
Further, most every forensic case involves a variety of different specialised personnel, with both scientists and law enforcement working closely together, and thus a mixture of skillsets. Addressing and removing ‘irrelevant context’ has therefore raised concerns that such an approach may create silos of different personnel that hamper an integrated approach within the practice of forensic science, and in the delivery of robust forensic reconstructions. In many cases it is acknowledged that contextual information will have a role in assisting in forensic reconstructions. Whilst not all contexts will have a biasing influence, experimental study 3 from this thesis does illustrate that it is possible for context to affect early hypothesis and decision-making, even with non-ambiguous evidence. It is therefore important to be aware of such instances, and to take steps to ensure that inferences are shielded as much as possible form potential cascaded bias from the exposure of context.
Finding an appropriate balance between the risk and benefits of enacting solutions that seek to deal with the issues of extraneous context is not an easy undertaking. It could be argued that ideally, the forensic anthropologist collecting the skeletal remains might need to be different from the analyst conducting the biological profile, in order to allow the analyst to carry out their assessment without context associated with the death scene or the body itself. This would mean a change in working practices that may not always be feasible or straightforward, but this approach has been successful in some laboratories within other disciplines (Kassin et al. 2013; Krane et al. 2008; Dror 2014a).