The hidden structure of fact-finding.

• Author: Spottswood, Mark
• Position: II. A Primer on Dual-Process Cognition C. System 2 Processes Information in a Conscious, Effortful Manner through Conclusion, with footnotes, p. 166-200

3. System 2 Processes Information in a Conscious, Effortful Manner

   Having considered System 1 processes in some depth, let us now turn our attention to System 2 cognition. One of the first key hallmarks of System 2 processing is that it takes place within the explicit view of conscious awareness and is thus available to introspective awareness. (145) When System 1 gives us an answer, we know what the answer is but not how we arrived at it. By contrast, when System 2 is doing the work, we are aware of the work that goes into producing the outcome and experience a series of internal steps involved in getting there. (146)

   A second, closely related hallmark of System 2 cognition is that it is ego depleting, meaning that we find it mentally taxing to employ and need to use willpower in order to keep it going. (147) All of us, I am sure, have experienced this feeling, such as when trying to do a complicated bit of mental math. (148) In this respect, System 2 cognitive processes seem analogous to physical exercise. People generally save their mental strength if they know they will have to use it in the future, and if they have recently expended their willpower to focus on a task, they will find it harder to repeat the performance until they have rested. (149) So, while our associative cognition happens automatically and effortlessly, our System 2 is activated only with an effort, and we will find ourselves weary if we use it for too long without taking a break.

   A third significant feature of System 2 is that it often involves the manipulation of words, symbols, or images held in working memory. (150) A bit of introspection shows that many of our conscious thoughts take linguistic or imagistic form. If we try to monitor the nature of our own thoughts, we may hear words in our mental ears or see images in our mind's eye. For some types of System 2 cognition, such sounds and images may be an integral part of mental processing. (151) For instance, when we add a column of numbers in our heads or consider the validity of a logical argument, we often hold symbols in mind in sequence, mentally applying rules in order to transform them in steps toward a final result.

   To be clear, the fact that symbols or mathematics is at stake is not a necessary sign of System 2 processing; if we are familiar enough with a particular problem, we may have integrated its answer into our underlying associative system. Thus, any adult who has learned his multiplication tables as a child knows the answer to the question, "What is 4 times 5?" without having to manipulate any numbers in active memory. But if the question shifts to, "What is 105 times 32?," no answer will spring to mind, and our associative system will be dumbfounded. Instead, we can solve the problem only by applying a series of learned rules, and, in so doing, we will experience all of the hallmarks of System 2 cognition. Our conscious minds will be occupied, and we will experience the task as effortful. It is this sort of symbolic manipulation that is another core, identifying feature of System 2 processing. System 1 can do many things well, but applying explicitly learned rules to truly novel situations is simply beyond its capabilities.

   One final benefit of System 2's ability to reason by way of the sequential manipulation of symbols and images is that it enables us to more easily consider hypothetical scenarios and their likely consequences. As we explored above, System 1's pattern recognition feature seems to rapidly assimilate even ambiguous stimuli to the closest known pattern. (152) Thus, once we have seen Lincoln's face in the squares, it is hard to see other potential patterns that they might make. This difficulty illustrates the general tendency of System 1 processing to default to a single model of reality. (153) But for many problems that we face--including many problems in the law the ability to reason counterfactually is critical. This ability requires simultaneously knowing the true state of the world and imagining what would happen if the facts were otherwise. System 2 provides us with a means to achieve this by letting us consciously manipulate symbols and images in order to engage in hypothetical reasoning. (154) Using these explicit semantic representations, we can describe events that we have not observed or which we do not currently believe and then imagine the consequences that would follow if our imagined scenarios were true, without as System 1 would--immediately treating our descriptions as the true state of affairs. Thus, this System 2 feature enables us to pose hypothetical or counterfactual questions and explore their answers.

4. The Complex Interplay of Dual-Process Reasoning

   Although the manipulation of symbols and hypothetical scenarios is an important function of System 2, it is not the whole story. A significant volume of System 2 activity involves the exercise of self-control, in which we try to guide or restrain the results of automatic System 1 processing. (155) If I tell you to stare intently at a particular word on a page for a long stretch of time, you will find that your conscious mind is occupied in an effortful way, indicating that System 2 is involved in the task. But this sort of effortful focusing of attention surely does not involve symbol manipulation. Similarly, if you have committed yourself to losing weight, you may sometimes find yourself in a common conundrum, in which a tasty treat is available to you but your goal is to avoid eating it. The common element in both settings is that we must resist our automatic impulses, which prompt us to pay attention to distractions and to eat anything that we associate with pleasurable tastes. And because System 2 resources require the active use of conscious willpower, we will find these innate impulses harder to resist as we become distracted or tired.

   This self-control function, combined with System 2's potential to use semantic reasoning to avoid associatively driven errors, is potentially very attractive as a means of improving the judgments and choices that would arise from an umnonitored System 1. Thus, for instance, there is a well-documented feature of Type 1 processing known as "belief bias," in which people who are trying to work out whether a logical argument is technically correct will often approve of faulty arguments if those arguments are used to support true conclusions. (156) It turns out, however, that the extent to which people fall prey to belief bias hinges on the degree to which System 2 processes are actively involved in solving the problem. Thus, when research participants must allocate working memory to other tasks or answer too quickly to think through the problem sequentially, errors are more likely; conversely, when people are encouraged to reason deductively, such errors are reduced. (157)

   This potential for these conscious overrides of System 1 responses suggests that a good way to avoid System 1 errors would be to employ System 2 reasoning as often as possible. This approach, however, is not the panacea it may seem. System 2 processing is not always engaged in order to monitor or override our intuitive impulses. It can also be engaged to defend or explain such instincts in ways that make them appear more reasonable than they really are. (158)

   The role of System 2 as an unwitting apologist for the results of unconscious cognition has been well documented. One type of evidence comes from studies of "split-brain" patients, who underwent surgeries to sever the fibers that connected the two hemispheres of their brains (the corpus callosum) in order to treat chronic seizures. (159) The split-brain patients allowed researchers an unusual opportunity because the left hemispheres of their brains, which typically manage language functions, had no direct inputs from the right hemispheres, which lack speech-producing capabilities. In studies of such patients, researchers have observed a surprising phenomenon: the left, speaking side of the brain will invent reasons for actions initiated by the right, silent side, even though the former is ignorant of the stimuli that motivated the latter's action. (160) Even more strikingly, patients who engage in such confabulation do not seem to be aware that their fabricated reasons were not the real reason for their actions. (161)

   This sort of confabulation is not confined to split-brain patients, however. One striking example of System 2's ability to generate false reasons for unconsciously induced choices was seen in an experiment by Richard E. Nisbett and Timothy DeCamp Wilson, who arranged four identical pairs of nylon stockings on a table and then asked passersby to give them comparative ratings as part of a fictitious marketing survey. Despite the fact that the stockings were identical, viewers expressed a clear statistical tendency to prefer the rightmost item (chosen by 40%) and to disfavor the leftmost item (12%). (162) The exact reasons for this "right-side bias" are somewhat mysterious, (163) but the main point is that none of the participants believed that the location of the stockings had anything to do with their preferences, and all but one participant (who happened to be a psychology student) denied that any such influence was possible when asked directly. Rather, they invented fictitious differences between the stockings in terms of "knit, weave, sheerness, elasticity or workmanship" to justify their right-skewing preferences. (164)

   One final example may serve to drive the point home as starkly as possible. Thalia Wheatley and Jonathan Haidt used hypnosis to prime participants to feel disgust when reading certain words and then had the participants evaluate several stories. (165) One of these stories involved no plausible moral violation: it described a student council leader who was "in charge of scheduling discussions about academic issues" and who tried to choose topics that "appeal to both professors...