Imagination and Planning

[Itzhak Gilboa]

Hi all,

We've just completed a (tentative version of a) paper on (case-based) planning and imagination.  The abstract is below and the paper is available at

https://itzhakgilboa.weebly.com/uploads/8/3/6/3/8363317/gayer_gilboa_imagination_and_planning.pdf

Any comments are very welcome, of course.

Happy new year!

Gabi and Tzachi

_________________________________

We consider a model of case-based planning, where a position is a vector of numbers, and a case is an edge in the directed graph of positions. The planner generates new plans by using cases that are similar to those she has observed in the past. In the benchmark model presented here, similarity is defined by equality of differences (between the target and the source position). We prove a complexity result that shows why planning requires imagination and is not easily done algorithmically. We put this result in the context of learning and expertise in case-based models, distinguishing among information, insight, and imagination.

[Itzhak Gilboa]

Hi,

We have just published a brief report in PNAS titled "Optimizing the Application Order under Precedent-Based Decision-Making".

The abstract is below and the paper is available at 

https://itzhakgilboa.weebly.com/uploads/8/3/6/3/8363317/ag_optimizing_application_order_pnas_2025_with_si.pdf

All the best,

Rossella and Tzachi

__________________________

We study the decision problem of a Proposer who has a set of applications to submit for approval to an Authority and can choose an order of submission. The Proposer’s utility depends on the Authority’s rulings. The Authority has to be consistent with its past decisions, which we model using the nearest-neighbor criterion. If the Proposer’s utility increases with the set of approved applications, then any greedy strategy is optimal for her: She should submit any application that, given the current history, would be approved. However, if her utility increases with some approvals but decreases with others, the Proposer’s problem becomes significantly more complex. In the single-dimensional case, an optimal strategy can be computed in polynomial time. In the general case, however, finding an optimal strategy is NP-hard. Thus, even in the absence of uncertainty or strategic behavior on the part of the Authority, evaluating the impact of current submissions on future outcomes can be computationally intractable.