This post is part 3 of The downbeat technique, a series looking at a physical analogy to exit writer's block, fatalism, and burnout at a core psychic level. In part 1, I discussed the motivation behind it, in part 2 I went through how to use it effectively as a software engineer, and here I review a neat computer science education paper on the cognitive science of avoidance.
Persisting cognitive dissonance + high cognitive load = avoidance failure
I searched a few of the terms I was thinking about (namely "avoidant psychology software engineers") and found what I believe is a paper matching the causal mechanism I intuited from experience. Fascinatingly, it focused on the concept of cognitive dissonance (hereafter CD) rather than what I had been vaguely calling "discomfort", which is more precise and gratifying to stumble upon.
In the paper, Reducing Persisting Cognitive Dissonance and Dropout Rates in Computer Science, the author discusses the traditional learning model of constructivism with an eye fixed on the high cognitive load produced.
It picks out metacognitive skill set as a limiting factor, and unresolved conflicts and persisting cognitive dissonance as the specific causes of failure.
The combination of cognitive dissonance and high cognitive load taxes working memory and prevents conflict resolution
In a computer science context this might mean unresolved bugs in a computer program.
The obvious solution is to reduce CD (e.g. by positive framing) and improving metacognition. The author suggests environmental scaffolds (the technical artifact is a debugging IDE essentially) to support positive outcomes.
The table of contents gives the outline of the thesis:
- The technical artifact aims to protect users (i.e. programmers) from persisting CD.
- It discusses cognitive psychology, as well as a sociological perspective, behaviorism, and CD.
- §2.2.3 goes into depth on CD:
- Dissonance is caused by inconsistencies in cognition.
- New information causes this CD, i.e. produces inconsistencies, and a person actively hides the causes of CD.
- This is somewhat contradictory because CD is not maintainable.
- Reducing it requires resources, much like tech debt is paid down.
- However, the cost of reducing CD depends on a person.
- When allowed to persist, CD leads to avoidance behavior, which is a useful but dangerous strategy and can lead to what the author calls entrapment.
- Instead, the author suggests focus to process instead of results to reduce avoidance.
- §2.2.3.10 connects CD, motivation, goals and positive reinforcement.
- §2.2.4 discusses loss aversion and prospect theory:
- §2.2.4.1: perception of risk leads to poor performance,
- §2.2.4.2: risk-taking is a warning sign.
- §2.2.4.3: the framing effect – positive expressions improve motivation.
- Subsequent sections cover memory itself (§2.3) and learning (§2.4).
- §2.3.1.2: chunks and schemas circumvent working memory capacity limit
- §2.3.1.3: element interactivity is introduced as the cause of CL.
- §2.3.1.4: high CL compounds damage from CD, causing avoidance.
- §2.4.1.3: improved metacognition improves debugging, a.k.a. problem solving.
- §2.4.1.4: metacognitive skills can be improved by intelligent tutoring systems.
- Help-seeking is prevented by fear of losing face, so anything that reduces that fear and increases access to help will of course also reduce the failure rate.
- §2.4.4: frames bugs as solutions instead of problems.
- §3: conclusions on the requirements for a system using these techniques to assist workers.
Some other interesting things to note:
- §3.1.2: deadlines can induce beneficial CD.
- §3.2.4: deadlines for large or difficult tasks can be harmful.
- §3.1.5: metacognition involves "knowing what to expect".
- §3.2.7.2: graceful failure prevents complete failure and reduces damage.
Conclusions on the paper
Previously, I'd been considering the working of my approach as being spelling out the cognitively dissonant aspects of a task, and then conversely focusing on how to 'harmonise' that dissonance.
In this thesis, CD is considered to be initiated by the receipt of new information, rather than being an existing aspect of a task. This means that CD occurs when we create, receive, or think of new information. As a result, our definition of when to "think dissonantly" can be relaxed from an "always on" mode to one that incorporates CD alongside new information.
It likewise lets us relax our definition of when to stop thinking dissonantly (or when to accept CD) by seeing it as something that we reduce when we have the available resources. This suggests CD is harmful when we do not plan to reduce it, and directly connects the "local" (problem level) CD avoidance with the "global" (task level) avoidance failure that comes when unaddressed CD jeapordises success.