As an anthropologist, I find academic culture fascinatingly full of contradictions. Perhaps that is why, although I find this target article convincing in its details, I interpret its significance rather differently.
As I understand it, Elqayam & Evans (E&E) distinguish between a “directive ought,” in which the statement “one should x” is conditional upon the desire to achieve a specific goal, and an “evaluative ought.” The authors write: “It is only when formal systems are regarded as having a priori, unconditional value, that the ‘ought’ becomes an evaluative one” (sect. 2, para. 5). The absolute, unquestioned form, it seems, can lead one towards the academic sin of bias.
My first concern is this: In practice, the distinction between “oughts” is fuzzy. For example, does the researcher have to be personally conscious of the goal-directed rationale behind a disciplinary norm to avoid an evaluative ought, or is it enough that it exists? For example, in anthropology there is a clear bias against working for the military. Our disciplinary history shows why we would be cautious, but it would seem the bias is more widely known than the specific reasons for it. Similarly, must the goal be made explicit to avoid the accusation of evaluative normativism?
Many of my immediate colleagues employ evolutionary perspectives. We often casually refer to the “purpose” of a trait, or the “reason” it was selected for. We know evolution is not purposeful – but a third-party observer could be forgiven for perceiving a teleological error on our part. Perhaps we should not interpret too deeply into researchers' choice of words when they do not specifically address the meaning of “ought” in their work.
For example, E&E hold Gigerenzer and Todd (Gigerenzer et al. Reference Gigerenzer and Todd1999) as an example of the proper way to avoid evaluative normativism. Gigerenzer, they say, “famously repudiates normative systems, arguing that heuristic rules of thumb outperform normative computations” (target article, sect. 1, para. 8; cf. Gigerenzer Reference Gigerenzer2007). To my mind, Gigerenzer and Todd do not do away with normativism but rather substitute one norm with another: they clearly think that problem-solving “ought” to be cost-effective (i.e., considering not just accuracy but also effort and the costs of error).
If we stipulate that academia is a problem-solving institution, perhaps we should consider normativism using a similar cost/benefit analyses. How hard would it be to eradicate evaluative normativism? And when it expresses as bias, is it a “bad thing,” or could there be positive effects?
We should consider how deeply ingrained normative judgments are in the process of academic progress – even in the assumption that progress is “a good thing.” It is somewhat ironic that E&E appear to be making a normative claim when they say we should avoid normativism. One could object that their claim is directive – that this “ought” implies a reason: to further the scientific understanding of human behavior. But this merely points towards another a priori assumption: that understanding human behavior, and specifically understanding it in a particular way, is a “good thing.” There would seem to be a strong interaction between our evaluative “oughts” and academic progress (cf. Kuhn Reference Kuhn1996). E&E say that normative research biases affect what researchers choose to study (sect. 6, para. 9); one could also say, normative research biases concentrate efforts. Perhaps progress can be described in terms of improved (or, at least, more useful) normative assumptions.
But I would make two stronger claims. First, bias may be a potentially useful part of academic method. Consider peer review, one of the primary evaluatory mechanisms that allows academia to function. Much effort is expended into making peer review “unbiased,” for example, by hiding authors' identities from reviewers, lest they fall prey to the “Matthew” and “Matilda” effects that bias reviews in favor of high-status (Merton Reference Merton1968) and male (Rossiter Reference Rossiter1993) academics. While double-blinding conforms to the academic norm of objectivity, it also saves us from having to decide when sex and/or status biases can be justified. For example, it may be useful to allow a bias that favors women in fields where they are underrepresented (e.g., engineering, philosophy).
Second, our extra-academic a priori norms may alert us to potential conflicts between the intellectual merit of a practice within academia and the broader impacts on society. (My reference to the National Science Foundation's evaluative criteria is deliberate.) For example, one might feel that the President of Harvard ought not to suggest that women may be less represented in the sciences due to innate differences in their mathematical ability (Bombardieri Reference Bombardieri2005); or find it inappropriate that the excavation of the Manhattan antebellum African American Burial Ground began without significant African American participation (Harrington Reference Harrington1993). In either case, regardless of our evaluation of intellectual merit, our extra-academic norms alert us to possible negative impacts.
In summary, E&E make what seems to be a valid point regarding a particular normative assumption. I would suggest that we not generalize from this, but consider all biases – even ones that produce gross biases – individually, in light of their impact on academia and more broadly. Do as E&E have done, and make the a priori assumptions explicit, evaluate the biases they produce, and examine available alternatives instead of dismissing even evaluative normativism out of hand.
You have
Access
As an anthropologist, I find academic culture fascinatingly full of contradictions. Perhaps that is why, although I find this target article convincing in its details, I interpret its significance rather differently.
As I understand it, Elqayam & Evans (E&E) distinguish between a “directive ought,” in which the statement “one should x” is conditional upon the desire to achieve a specific goal, and an “evaluative ought.” The authors write: “It is only when formal systems are regarded as having a priori, unconditional value, that the ‘ought’ becomes an evaluative one” (sect. 2, para. 5). The absolute, unquestioned form, it seems, can lead one towards the academic sin of bias.
My first concern is this: In practice, the distinction between “oughts” is fuzzy. For example, does the researcher have to be personally conscious of the goal-directed rationale behind a disciplinary norm to avoid an evaluative ought, or is it enough that it exists? For example, in anthropology there is a clear bias against working for the military. Our disciplinary history shows why we would be cautious, but it would seem the bias is more widely known than the specific reasons for it. Similarly, must the goal be made explicit to avoid the accusation of evaluative normativism?
Many of my immediate colleagues employ evolutionary perspectives. We often casually refer to the “purpose” of a trait, or the “reason” it was selected for. We know evolution is not purposeful – but a third-party observer could be forgiven for perceiving a teleological error on our part. Perhaps we should not interpret too deeply into researchers' choice of words when they do not specifically address the meaning of “ought” in their work.
For example, E&E hold Gigerenzer and Todd (Gigerenzer et al. Reference Gigerenzer and Todd1999) as an example of the proper way to avoid evaluative normativism. Gigerenzer, they say, “famously repudiates normative systems, arguing that heuristic rules of thumb outperform normative computations” (target article, sect. 1, para. 8; cf. Gigerenzer Reference Gigerenzer2007). To my mind, Gigerenzer and Todd do not do away with normativism but rather substitute one norm with another: they clearly think that problem-solving “ought” to be cost-effective (i.e., considering not just accuracy but also effort and the costs of error).
If we stipulate that academia is a problem-solving institution, perhaps we should consider normativism using a similar cost/benefit analyses. How hard would it be to eradicate evaluative normativism? And when it expresses as bias, is it a “bad thing,” or could there be positive effects?
We should consider how deeply ingrained normative judgments are in the process of academic progress – even in the assumption that progress is “a good thing.” It is somewhat ironic that E&E appear to be making a normative claim when they say we should avoid normativism. One could object that their claim is directive – that this “ought” implies a reason: to further the scientific understanding of human behavior. But this merely points towards another a priori assumption: that understanding human behavior, and specifically understanding it in a particular way, is a “good thing.” There would seem to be a strong interaction between our evaluative “oughts” and academic progress (cf. Kuhn Reference Kuhn1996). E&E say that normative research biases affect what researchers choose to study (sect. 6, para. 9); one could also say, normative research biases concentrate efforts. Perhaps progress can be described in terms of improved (or, at least, more useful) normative assumptions.
But I would make two stronger claims. First, bias may be a potentially useful part of academic method. Consider peer review, one of the primary evaluatory mechanisms that allows academia to function. Much effort is expended into making peer review “unbiased,” for example, by hiding authors' identities from reviewers, lest they fall prey to the “Matthew” and “Matilda” effects that bias reviews in favor of high-status (Merton Reference Merton1968) and male (Rossiter Reference Rossiter1993) academics. While double-blinding conforms to the academic norm of objectivity, it also saves us from having to decide when sex and/or status biases can be justified. For example, it may be useful to allow a bias that favors women in fields where they are underrepresented (e.g., engineering, philosophy).
Second, our extra-academic a priori norms may alert us to potential conflicts between the intellectual merit of a practice within academia and the broader impacts on society. (My reference to the National Science Foundation's evaluative criteria is deliberate.) For example, one might feel that the President of Harvard ought not to suggest that women may be less represented in the sciences due to innate differences in their mathematical ability (Bombardieri Reference Bombardieri2005); or find it inappropriate that the excavation of the Manhattan antebellum African American Burial Ground began without significant African American participation (Harrington Reference Harrington1993). In either case, regardless of our evaluation of intellectual merit, our extra-academic norms alert us to possible negative impacts.
In summary, E&E make what seems to be a valid point regarding a particular normative assumption. I would suggest that we not generalize from this, but consider all biases – even ones that produce gross biases – individually, in light of their impact on academia and more broadly. Do as E&E have done, and make the a priori assumptions explicit, evaluate the biases they produce, and examine available alternatives instead of dismissing even evaluative normativism out of hand.