| 总结: | <p>It would be a daunting task to learn everything from scratch each time we are faced with a new problem. Instead, humans and other animals are helped by the fact that few situations in life are truly novel. Thus, we can draw upon previous experiences of similar situations and come up with our ‘best guess’ of how to deal with a new task at hand. In this thesis, I will attempt to answer the question of how this ability to generalise knowledge between different situations might be achieved in the brain.</p>
<p>To be able to apply previously learnt knowledge to new but related situations, the common structure needs to be abstracted away from the sensorimotor specifics of experience. In my first set of experiments, I showed that mice, like humans, can learn to transfer knowledge between different sensory scenarios, and get better at solving each new sensory example of a problem. A series of behavioural analyses I conducted suggest that animals can abstract strategies away from the specifics on various time scales and could not only learn to generalise the short-term abstract sequences of events, but also a long-term strategy that integrated the recent history of animals’ experiences.</p>
<p>In the second set of experiments, I focused on how these abstractions might be achieved in the brain, and how abstract representations might be tied to the sensorimotor specifics of each new situation. I found that neurons in medial prefrontal cortex (mPFC) maintained similar representations across multiple problems, despite their different sensorimotor correlates, whereas hippocampal (dCA1) representations were more strongly influenced by the specifics of each task. These data suggest that mPFC and hippocampus play complementary roles in generalisation of knowledge, with the former abstracting the common structure among related tasks, and the latter mapping this structure onto the specifics of the current situation.</p>
<p>Together, these experiments aim to develop our understanding of a hallmark of biological intelligence – how prior experience is leveraged to solve novel, real-world problems.</p>
|
|---|