> **Learning** is the process of acquiring new understanding, knowledge, behaviours, skills, values, attitudes, and preferences. ## Relevance to Risk Management Learning is necessary to improve decision-making. - [[Risk Aware Culture]] - [[Continuous Improvement]] ## Learning through experience Humans learn through processing embodied experiences. Sensory data such as vision, audio, and physical touch is processed by the body, including the brain. Many cognitive scientists now model human learning from a [[Bayesian]] perspective. In a bayesian model, humans update their prior beliefs (expectations) with new data (evidence) as they have new experiences. The new probability distribution describing their beliefs is called the posterior. ![[Pasted image 20240827214055.png]] ## Analogical Reasoning An analogy is a likeness perceived between things, like A and B. > "Petal is to flower as leaf is to tree." > > *ChatGPT (2024-12-25)* This is an example of a "Part to Whole" analogy. A petal is *like* a leaf, because it's similarly part of a larger plant. **Analogical reasoning** is the process of *inference* based on *reasoned similarities* between two or more entities. If A and B are alike in some way, they're likely to be alike in other ways. Using analogical reasoning, a learner infers other shared characteristics of A and B. To return to leaves and petals; > "Contribution to Purpose: > - Petals often attract pollinators, aiding reproduction for the flower. Leaves are essential for photosynthesis, providing energy for the tree. > - **Reasoning**: Both petals and leaves are vital for the survival and reproduction of their respective plants, though they serve different roles." > > *ChatGPT (2024-12-25)* ## Machine Learning Generating reasonable [[Chains of Thought (CoT)]] through deliberate analysis of how things are *alike* and how they are *different*. The [[O1, O3 Systems (OpenAI)]] of models implements reasoning by training on logical "chains of thought". These thoughts may be generated with GPT-4, which is called "Synthetic Data".