When I was in high school, it was generally thought that agriculture drove the development of towns: that permanent settlements were not useful without farming. The first major towns known at that time dated to around 5000 BC.
Discoveries of new sites and further investigation of known sites in the 1990s upended this view somewhat. Sites such as Hasankeyf Höyük, Göbekli Tepe, Tell es Sultan and Tell Abu Hureyra were founded some 12000 years ago, with permanent stone structure but no sign of agriculture or pottery in their earliest layers. Farming at these sites came over 1000 years later: driven, or at least facilitated, by the permanent settlement.
The motivation for hunter-gatherers to build settlements would be basically the same as the motivation for them to make repeated use of natural caves. Some areas provide well enough that a living could be made without leaving a small range. Some peoples may have a need for a winter base to tuck away during the low-hunting season. And perhaps it just became a pain in the arse continually rebuilding new settlements and lugging all of your tools from place to place, as tool cultures became more specialised and complex.
These sites represent the peak of the Natufian culture which commenced around 13000 BC. Early in that period, people began to build semi-permanent villages of a handful of people that had long term periodic use.
https://www.jstor.org/stable/40024495
On ‘Sedentism’ in the Later Epipalaeolithic (Natufian) Levant
https://onlinelibrary.wiley.com/doi/abs/10.1002/arp.1777
A Neolithic sedentary hunter–gatherer settlement with densely arranged buildings: results of geophysical prospection at Hasankeyf Höyük in south-eastern Anatolia
https://youtu.be/iSG1MsQSo_A
Gobekli Tepe, dawn of civilisation
