Weekly Roundup: More on the Political Brain, Biased AI, New Neurons for AI, and New Developments on Waste Disposal in the Brain

Let’s start this week by revisiting a topic I posted on this week namely on political brains before moving on to a number of recent studies out on AI as another very topical theme. We can then close this round up out by looking at cognitive function and waste disposal in the brain.

---

More on the Political Brain

Just this week I reported on how those with lower self-awareness have higher brain emotional responses to political topics that they perceive as moral.

---

Research Hit: Low Self-Awareness Increases Emotional and Moral Brain Responses to Political Topics

---

This study just out by Siddiqi et al. of Northwestern University identified specific brain networks that are associated with intensity pf political engagement by looking at those with localised brain injury. In this cohort it was Vietnam war veterans (the analysis was originally done a while ago now).

This differs to many studies nowadays that aims to identify activation patterns in healthy brains when engaging with specific tasks or decision making. The study also tried to evaluate any changes in politicising before or after localised brain injury.

Like many studies previously, they did not find any region or circuit that was related to a political ideology, but they did identify regions that were associated with how engaged or responsive the veterans were to political content.

So what regions engage with political content?

Two general regions that are well-known in and out of neuroscience circles seemed to influence this political engagement.

1. Prefrontal Cortex damage increased the intensity of engagement. The prefrontal is involved in cognitive control so this reduction in control leads to an increase in intensity.

2. Amygdala damage lowers the emotional response. The amygdala is considered an emotional centre and is involved heavily in fear and threat responses and therefore damage here logically reduced this emotionality.

This would fall in line with what we may observe in political discourse - an inability to control and engage cognitively with political topics and high emotionality. If either of those are higher or lower, how you engage with political discourse will change.

But maybe AI can save use by giving us unbiased decisions and information - if we are willing to listen to AI in the first place that is. Some research has shown that AI can convince political partisans or conspiracy theorists much better than human beings can (probably because AI is endlessly patient).

Well, maybe not, according to another piece just published.

---

Biased AI - Just Like Us Humans

Chen et al. of Western University in Canada put ChatGPT through 18 different bias tests. I am just writing about bias and decision making in the Handbook of the Brain in Business because these are decision making traps that us human beings fall into consistently. These include things like the framing bias, how you say things, or focus on information, loss aversion in which we prioritise losing things over winning things and 16 other biases.

You my assume that ChatGPT would be well-trained to avoid these biases and be perfectly rational considering the amount of data it uses to come to its output. But according to these researchers not so.