Maximising expected outcomes – framed as expected utility in behavioural science and expected reward in reinforcement learning – has long dominated models of decision-making. In this seminar, Ollie Hulme (DRCMR) asks if people maximise time averages instead. He will present data showing that time average models better account for both neural and behavioural data in risky choice experiments, with participants operating surprisingly close to the theoretical optimum. Ollie will also discuss arguments for the biological and cognitive plausibility of time averaging as well as its consilience with a diversity of theories in psychology and neuroscience.
The seminar was hosted by Emilie Rosenlund Soysal (London Mathematical Laboratory).
Regarding Ole’s comment in the end of the seminar: indeed, he was referring to the experiments by Arne Vanhoyweghen on gambles with lower and upper boundaries. Read more in Chapter 7 of Arne’s PhD thesis: https://researchportal.vub.be/en/publications/ergodicity-and-human-decision-making-on-the-impact-of-path-depend/
I have a question, perhaps naïve, but I’ll still ask it.
What if, _a posteriori_, we can think that the biggest confirmation of the results of the neurological experience is the widespread observation of greed in the world?
Let me put it this way: if the reward is pleasant but damped as the absolute value goes up in size, then it is to be expected that the more you have, the more you want.
If the spike you get for a gain stays the same because of the logarithmic utility function being sort of “hard-wired” into our nervous system, as the scan shows, then no wonder that we observe this dynamics in real life.
Conversely, if the spike went up at the same rate as the absolute value of the wealth gain, then it would likely become unbearable after a while, because a finite nervous system can accommodate an increasing signal only up to a (finite) point.
This “remark” of mine has no moral underpinning whatsoever and I totally understand that this is not a rigorous statement, but a hand-waving connection. But I still wonder if Oliver Hulme or anybody else would subscribe to a tentative extrapolation of his group’s results of this tone.
ERRATA: I did not mean, of course, “the spike you get for a gain stays the same”, but “the spike you get for a gain does not grow that much with absolute value”…I am referring, of course, to the blood oxygen spike that the MRI measures.
Apologies for the misstatement above…
Perhaps the greed is baked in more deeply into the theory which is the starting assumption that people have the objective of maximising the growth rate of wealth (or whatever it is they care about growing). If that were the case, then regardless of the dynamics, and regardless of the transformation being imposed to achieve the growth, people would always be seeking to maximise the growth rate, and that implies there is never enough. Of course this an idealised assumption, and human motivation is much more complex than this. So i would say “greed” is a deeper assumption that is actually independent of the dynamics, and independent of whether the transformation is logarithmic or not.