IFL Science! Is one of my favorite web sites. Although its writers can sometimes be found sliding down the slippery slope between scientific journals and popular media, they at least tend to slide a bit more slowly than journalists who are mainly interested in being entertaining. A recent post on this site titled “Researchers observe the exact moment when a mind is changed” piqued my interest, given the considerable challenge of measuring change in a non-physical “phenomenon.” (Placing “phenomenon” in quotes should suggest that I’m uncomfortable using this term when its referent has no physical features.)
Yet another “how the brain explains behavior” study
The IFL Science! post reported that Stanford University researchers trained monkeys to perform “decision-making tasks” while their brain activity was being monitored. In this project, two rhesus macaques were trained to trace a finger through a maze that led to one of two targets. On some trials, both targets were available, but on others one choice was blocked, leaving only the alternative target for a response. The researchers argued that they could make this switch while the monkey was still making a decision, thereby encouraging the monkey to change its mind.
To do this, the researchers concurrently measured the electrical activity of two brain regions involved in the “planning” of movement (more quotes). They proposed that this enabled them to reliably predict the selected target a fraction of a second before the monkeys were prompted to move their finger. One of the researchers was quoted as saying, “We saw that the brain activity for a typical free choice looked just like it did for a forced choice. But a few of the free choices were different. Occasionally, [the monkey] was indecisive for a moment before he made any plan at all. About one time in eight, he made a plan quickly but spontaneously changed his mind a moment later.”
Except the conclusions are all about the mind
There would surely be more than a few methodological details we might take issue with were we to pour over the original publication in eLife, but they are beside the point for this discussion. Let’s instead focus on the interpretation of the animals’ behavior – at least that’s what behavior analysts would be interested in. The researchers were more interested in the mental processes taken as explaining their behavior, however, a priority as old as psychology and unhesitatingly accepted by biologically inclined colleagues.
Of course, the researchers were not looking at mental activity, but at measures of the electrical activity of part of the brain to the extent revealed by their measurement technology. This is an important clarification, given that their comments imply that they were observing decision-making. Their argument was that their procedures allowed them to “track single decisions.” Don’t miss the fact that the sole basis for these inferences about “free choices,” “plans,” and “changing minds” were variations in patterns of selected electrical activity across different types of trials in a fraction of the second before the animals moved a finger one way or the other. The implicit proposal that the direct measures of brain activity served as indirect measures of “mental events” is the glaring mentalistic faux pas, though perhaps it’s only flagrant to behavior analysts and others who are uncomfortable with philosophical dualism.
It might have been slightly less problematic had the investigators’ mentalistic interests been couched in more cautious terminology. Instead, they boldly described brain activity data in colloquial terms – as measures of decision-making, planning, or mind changing. This slippage exacerbates the risk that colloquial dialect is driving the science, rather than the other way around. That is, reifying everyday terms encourages researchers to try to explain the mental “phenomena” implied by everyday dialect, rather than to discover genuinely revealing biological phenomena associated with behavior and describing the results with scientifically-based terminology. In the researchers’ approach, the social contingencies underlying vernacular dialect set the scientific agenda, rather allowing it to be shaped by the emerging discovery of the interface between behavior and biology.
It’s about the implications
Behavior analysts know that our descriptions of human affairs are entirely learned. We learn to describe human behavior under certain circumstances as “planning,” we are taught to say that we are “making a decision,” and we come to say that we have “changed our mind” only after others have modeled this construction. This vocabulary works well enough for ordinary discourse, but mentalistic phrasings such as these are burdened with implications that make them a poor guide for coming up with scientific questions, developing experimental procedures, and interpreting data.
One such implication associated with everyday dialect is that its references to mental qualities require the assumption that these “mental events” somehow determine behavior, a causal relationship routinely accepted in behavioral neuroscience. In the present study, the measures of brain activity of interest were described as occurring prior to the target responses – well, hundreds of milliseconds prior anyway – and were taken as constituting the biological representation of these mental “events,” as well as the evidence for their causing the “choice” of one response or the other.
It’s important to be clear that no evidence is available from the study about this presumed causal relationship other than correlated changes in measures of electrical activity in a part of the brain. The fact that the changes of interest in these measures occurred only hundreds of milliseconds before a choice response is implicitly taken as a meaningful interval. The changes occurred (though only about one time out of eight opportunities) just barely before responses, encouraging the assumption of a familiar form of causation. The fact that the responses occurred with such a short latency after the changes in brain activity is consistent with the assumption that mental “events” such as choosing or making a decision are exceedingly brief and require little time to execute. Once these definitions are assigned and these relationships are assumed, evidence of the sort collected in this study is easily accepted as causal. Ergo, the decision-making (represented by the electrical measures) caused the responses.
The focus on changes in brain activity that unpredictably occurred on only a few trials and only hundreds of milliseconds before target responses almost sounds as if the researchers were stretching to reach their conclusions. This kind of reasoning seems the rule in behavioral neuroscience, however, where it is viewed as straightforward and reasonable. If you start by focusing on putative mental phenomena such as planning, choosing, and making a decision – and if these qualities do not actually exist as distinct physical phenomena – there is no alternative to proposing that they can be inferred from observed behavior and indirectly observed through directly measured biological phenomena. Convincing those who are more than a little bit suspicious that such “mental events” exist and function as assumed has proven exceedingly difficult. (See Uttal, 2011.)
It’s not that the biological underpinnings of behavior are an inappropriate scientific curiosity. Absent the gratuitous assumptions about mental “phenomena,” the reported results are potentially interesting, possibly even in a practical way. (The article suggests that they may aid the development of prosthetic arms controlled by the patient’s brain.) If procedures could be developed to more reliably isolate a unique pattern of brain activity that corresponded to aspects of choice behavior, it might be worth studying the environmental conditions under which that pattern and associated responding were correlated. Such a relationship would not require invention of mental qualities or a priori assumptions about causation.
Uttal, W. R. (2011). Mind and brain: A critical appraisal of cognitive neuroscience. Cambridge, MA: MIT Press.