Among many things, the man is terribly funny and doesn't hold back stirring up the Chomskyists 😂

Thrilling to attend a lecture by Geoffrey Hinton at the Royal Institution this weekend about the development and future of AI - with a reminder of the risks (arxiv.org/pdf/2310.17688)

Further proof of the existence of this ‘mood bias’ comes from RT (the costly option gets chosen faster when happy & slower when sad) and in gaze patterns (the costly option is inspected faster when happy; the uncostly option when sad). [6/9]

It was particularly exciting that we could ‘reconstruct’ mood purely with physiological valence and arousal markers. So, emotion ratings were not even necessary! Importantly, *physiological* mood bias and the *rated* mood bias were correlated. [5/9]

When modelling this effect, it could best be explained as follows: mood modulates a choice bias towards the costly but more rewarding option. We reproduced this finding in a confirmatory study. [4/9]

We found emotions on the mood continuum (happiness-neutral-sadness) to be associated with changes in economic choice. When happy, people sought larger rewards (in exchange for risk/delay/effort). When sad, people accepted smaller rewards that were safe/immediate/effortless. [3/9]

With text vignettes and music, we induced short (<1 min) states of happiness, sadness, anger, and fear. We validate the successful inductions with self-report and physiology: pupil dilation, skin conductance, facial musculature of the zygomaticus (😊 ) and corrugator (☹) [2/9]