²³Alternatively, the more sophisticated theory of reinforcement learning may provide another explanation even without such internal rewards. As explained in Chapter 5, in particular footnote 20, the RPE theory says there is frustration solely created by predictions in case you move to a state of lower value, and there is no reward. This could be the case when faced with a tiger if we assume that at every time step your probability of being eaten increases. Then, your chances of getting any positive reward in the future are getting smaller and smaller (because you won’t get any after being eaten), and thus the expected total reward during the rest of your life (which is the definition of state-value) is decreasing. The central assumption here is that your chances of being eaten are getting higher and higher in the presence of the tiger; this is clearly true at least in the beginning when you first perceive the tiger. However, it could be argued that such decrease in predicted reward does not really correspond to fear but something like disappointment.