27Selecting goals and successfully pursuing them in an uncertain and dynamic environment is an 28 important aspect of human behaviour. In order to decide which goal to pursue at what point in time, 29 one has to evaluate the consequences of one's actions over future time steps by forward planning. 30 However, when the goal is still temporally distant, detailed forward planning can be prohibitively 31 costly. One way to select actions at minimal computational costs is to use heuristics. It is an open 32 question how humans mix heuristics with forward planning to balance computational costs with goal 33 reaching performance. To test a hypothesis about dynamic mixing of heuristics with forward 34 planning, we used a novel stochastic sequential two-goal task. Comparing participants' decisions 35 with an optimal full planning agent, we found that at the early stages of goal-reaching sequences, in 36 which both goals are temporally distant and planning complexity is high, on average 42% (SD = 37 19%) of participants' choices deviated from the agent's optimal choices. Only towards the end of the 38 sequence, participant's behaviour converged to near optimal performance. Subsequent model-based 39 analyses showed that participants used heuristic preferences when the goal was temporally distant 40 and switched to forward planning when the goal was close.
41Author summary 42 When we pursue our goals, there is often a moment when we recognize that we did not make the 43 progress that we hoped for. What should we do now? Persevere to achieve the original goal, or 44 switch to another goal? Two features of real-world goal pursuit make these decisions particularly 45 complex. First, goals can lie far into an unpredictable future and second, there are many potential 46 goals to pursue. When potential goals are temporally distant, human decision makers cannot use an 47 exhaustive planning strategy, rendering simpler rules of thumb more appropriate. An important 48 question is how humans adjust the rule of thumb approach once they get closer to the goal. We 49 addressed this question using a novel sequential two-goal task and analysed the choice data using a 50 computational model which arbitrates between a rule of thumb and accurate planning. We found that 51 participants' decision making progressively improved as the goal came closer and that this 52 improvement was most likely caused by participants starting to plan ahead. 53 Introduction 54 Decisions of which goal to pursue at what point in time are central to everyday life [1-3]. Typically, 55in our dynamic environment, the outcomes of our decisions are stochastic and one cannot predict 56 with certainty whether a preferred goal can be reached. Often, our environment also presents 57 alternative goals that may be less preferred but can be reached with a higher probability than the 58 preferred goal. For example, when working towards a specific dream position in a career, it may turn 59 out after some time that the position is unlikely to be obtained, while another le...