%% ### #todo - [ ] Process the content in the "notes" section below. - [ ] Decide what to incorporate, then do so. - [ ] Move the rest of the notes to other notes, or delete it. - [ ] Re-publish this note. %% Learning is a complex, multifaceted topic that involves [[Learning involves myriad systems and processes within an individual that coordinate, interact, and unfold over time|myriad systems and processes that coordinate, interact, and unfold over time]].^[National Academies, 2018] Learning happens on [[What are the levels of analysis of learning?|many levels]], from the molecular level all the way to the societal level. Learning also happens on [[What are the timescales of learning?|many timescales]], from milliseconds to years and decades, and from birth through adulthood to death. Finally, learning happens in a [[To understanding learning, you need to consider of a wide variety of contexts and settings|wide variety of contexts and settings]]. Attempting to answer the question of what the mechanisms of learning are and how they work, then, is an ambitious undertaking. It's certainly beyond my capabilities to provide a satisfactory answer. Still, I'll try to take a few steps towards an initial, surface-level understanding of the matter. I decided to split my answer into three separate notes: - [[What are the biological mechanisms of learning and how do they work?]] - [[What are the cognitive mechanisms of learning and how do they work?]] - [[What are the sociocultural mechanisms of learning and how do they work?]] In the remainder of this note, I'll try to explain the rationale behind this organization. I'll start my exploration at the [[What are the biological mechanisms of learning and how do they work?|level of biology]], where my focus is on how learning works at the level of neurons, neural circuits, and brain regions. To me, this seems like the lowest level of analysis where we can talk about learning. The biology of the brain and its parts and interactions provides the [[The biology of the brain provides the physiological platform for learning|physiological platform for learning]].^[National Academies, 2018] Learning mechanisms at the biological level include changes in the synaptic strength between neurons, changes in the properties of individual neurons such as membrane excitability and dendritic branching, excitatory and inhibitory neurons in neural circuits, and the differentiation and integration of multiple distinct brain areas, to name but a few.^[E.g., Howard-Jones, 2018; Lieberman, 2021; Linden, 2019; Luo, 2021; Sapolsky, 2017] Next, at the [[What are the cognitive mechanisms of learning and how do they work?|cognitive level]], an array of mental processes enable the acquisition, processing, storage, and retrieval of information. Important mechanisms include sensation, which registers different aspects of the environment; attention, which selects certain sensory information for further processing; perception, which organizes, identifies, and interprets the attended sensory input; and memory, which encodes and stores information and from where it can later be retrieved and used.^[E.g., Bloom, 2023; Budson & Kensinger, 2023; Dehaene, 2021; Lieberman, 2021; Weinstein et al., 2019] Additionally, there are a number of metacognitive processes that monitor and regulate learning at this level.^[National Academies, 2018] Finally, there is the [[What are the sociocultural mechanisms of learning and how do they work?|sociocultural level]]. In humans, as well as many other animal species, learning predominantly occurs in a [[Learning is fundamentally social|social context]] and is influenced by [[Culture is the learned behavior of a group of people that is socially transmitted over generations and shaped to fit the group's circumstances and goals|culture]] — the values, beliefs, practices, norms, and tools shared by a group of people.^[Henrich, 2018; Hoppitt & Lala, 2013; National Academies, 2018] Throughout life, the individual learner interacts with and observes other people and their products, which creates a unique array of experiences that shape their attitude, skills, and knowledge and influence their behavior into the future.^[National Academies, 2018] Mechanisms at this level include, among others, the imitation of other people's behavior, observational learning of the consequences of other people's behavior, as well as coaching. Each of the above levels offers a different perspective on the learning process. In closing, I want to emphasize that these perspectives are utterly interconnected. If you look at a particular mechanism, say, at the cognitive level, you implicitly invoke all the mechanisms that preceded it as well as the ones that follow it.^[Sapolsky, 2017] None of these mechanisms work in isolation, but in concert. For example, a child learning a new skill might involve new neural connections (biological level), require the use of attention and memory (cognitive level), and be influenced by the attitudes and behaviors of parents and peers (sociocultural level). %% ### Notes - Ultimately, any learning seems to be the product of the forces of genes, culture, and individual experience (Bloom, 2023). - [[Psych]], chapter 5: - There is no learning (nurture) without initial machinery (nature). - [ ] However, this initial machinery typically requires some contact with the world (nurture) to work — no nature without nurture. For instance, vision needs input, else you go blind. - Still, sometimes it makes sense to say that something is learned (nurture) or innate (nature). - More generally, humans are evolved to acquire new information and new capacities. - Evidence from surgery where patients had parts of their brain removed might be a great way to illustrate the existence of multiple learning systems or mechanisms: - [[The Learning Brain]] offers a very concrete way to show that there are multiple types (or rather, mechanisms?) of learning: - Patient H.M. had parts of his brain removed to prevent seizures he was frequently experiencing. After the intervention, the seizures occurred less frequently. However, he wasn’t able to form new explicit (conscious, declarative) long-term memories anymore. At the same time, his ability to form implicit (non-conscious, procedural) long-term memories, and to hold information in working memory, was left intact. - HM’s story showed that human learning is not a unitary process. We have multiple learning systems that are different from each other and that depend on different brain systems. Explicit learning is different from implicit learning. Likewise, storing information in long-term memory works differently than storing information in working memory. - Seel (2012, p. 2695): For example, H.M. became amnestic after bilateral medial temporal lobe resection, a surgery that was performed to manage his uncontrollable seizure activity. After his surgery, he was unable to learn or remember any new sensory-perceptual or factual information. However, he was able to learn a variety of new motor and cognitive skills, including mirror tracing and solving the Tower of Hanoi. Nevertheless, he had no conscious memory of ever performing those tasks. Conversely, patients with Parkinson’s disease, in which the cortico-striatal system is primarily affected while the medial temporal lobe system remains intact, have great difficulty in learning mirror tracing and tower-type tests, while learning and memory dependent upon the medial temporal lobe system is pre- served (dependent upon the stage of the deteriorating disease process). - See also https://en.wikipedia.org/wiki/Implicit_learning#Amnesia_studies - From [[Books Do Furnish A Life]], p. 170 ff: It seems that the (or at least some) mechanisms of learning are genetically pre-programmed. Without them, a system can't learn anything at all. - Cf [[There's a continuum between determining a system's architecture at the start and letting the environment mold the system]] - Blank slates don't do anything. Unless there's some kind of set of motives, one can't have an intelligent system that does anything – that processes some inputs and ignores others, that treats certain inputs as reinforcers, etc. The system has to have some criterion for deciding when it is learning (e.g., what to pay attention to, how to analyze the input, what to be motivated by). If nothing else, that's got to be built in at the start. - Maybe add this point to the "there's a continuum" note above. Seems great. - A learning system has to have the machinery (i.e., the mechanisms) that allows it to learn. That machinery, I think, ultimately has to be explained in terms of evolution. - From [[Blueprint]]: - For **biologists**, the ultimate goal of genetics is to understand every path between inherited DNA differences and individual differences in behavioural traits, a **bottom-up approach**. - **Psychologists** focus on behaviour and use genetics to understand behaviour. This **top-down** psychological perspective begins with prediction. We can use inherited DNA differences to predict individual differences in psychological traits without knowing anything about the myriad pathways connecting genes and behaviour. - Maybe there's a similar distinction to be made here: You can either try to map out each mechanism in its entirety, or you can ignore most of the mechanism and simply focus on high-level correlations between inputs and outputs. - **General perspectives**: - ==**[[How People Learn II]], chapter 2 does a great job at highlighting the different forces that interact to shape learning** – e.g., the interplay of culture and biology. Maybe try to stay close to how HPL II discusses the issues; it seems like a good model to imitate.== - Especially the subsection "The dynamic interaction of culture, biology, and context" in chapter 2. - [[Learning involves myriad systems and processes within an individual that coordinate, interact, and unfold over time]] - [[Learning occurs within a dynamic system encompassing the changing contexts and people that surround an individual throughout life]] - Also the "conclusion" section in chapter 2. - E.g.: Learning is different for every individual because the interplay of their cultural, social, cognitive, and biological contexts. - [[Learning happens through action and experience in the world]] - [[The biology of the brain provides the physiological platform for learning]] - [[Dehaene's four pillars of learning]] – it seems like these are some very foundational learning mechanisms - **Domain-specific perspectives**: - See [[What perspectives on learning do different academic disciplines provide?]] - How to draw the distinctions between biology, cognition, and sociology? - [[Understanding How We Learn]]: "Cognitive psychology focuses on explanations related to the mind (e.g., encoding, consolidation, etc.), whereas neuroscience is concerned with figuring out what happens in the brain (e.g., physical activity in the brain)." - Should "biology" be understood in terms of the different levels at which parameters change discussed in [[Livewired]]? - Biology: [[What are the biological mechanisms of learning and how do they work?]] - [[How do the biological mechanisms of learning develop in an individual?]] - [[What is the evolutionary function of learning?]] - [[What is the evolutionary history of learning?]] - Cognition: [[What are the cognitive mechanisms of learning and how do they work?]] - Sociology: [[What are the sociocultural mechanisms of learning and how do they work?]] - Connect with [[What are the different types of learning?]]: To what extent do learning types make use of particular learning mechanisms? - [[Behave]] makes a strong argument against discipline-specific explanations, and that the categorization of science is mostly arbitrary. - Seel, 2012, neuropsychology of learning (p. 2459): "Everyday language lumps under the common label “learning” a wide range of activities of various levels of complexity, from learning our way through a new city to acquiring a first or a second language. It is unlikely that an assumption of general and common learning abilities underlies such a heterogeneous set of activities summarized as “learning.”" - Learning mechanisms operate as **semi-independent** and **self-organized systems**. - One challenge here is to map cognitive information processing to neural information processing. See [[International Handbook of the Learning Sciences]], chapter 7 for context. - [[Foundations of Embodied Learning]], chapter 5, distinguishes between three different types of processing: 1 = automatic and unconscious, 2 = explicit rule-based behavior, 3 = sociocultural interaction - [[How Learning Happens]], part V: - Learning is often seen as something **cognitive**, and that is definitely true. What we learn depends on how we process the information that we encounter and these processes take place in our brains. To learn, our brains must process new information and incorporate it into our existing knowledge or create new knowledge schemes. - But this is not the whole story. Learning is also a **social event** and thus the social environment – as with almost all our activities – also has a lot of influence on our learning. And that environment can greatly stimulate or stifle learning. %% --- Topics: - [[Learning (Index)]] Related notes: - [[What are the biological mechanisms of learning and how do they work?]] - [[What are the cognitive mechanisms of learning and how do they work?]] - [[What are the sociocultural mechanisms of learning and how do they work?]]