How do we learn by instruction and experience-- and how does negative negative feedback may influence our learning
Mattew Walch and John Anderson both scientists at the Department of Psychology at Carnegie Mellon, have corroborated through probabilistic experiments and modeling that humans and animals learn from trial-and-error interactions, experiences with the environment, but also through non-experiential situations that are recorded in the neural reward pathways.
The researchers made use of electroencephalograms (EGG), separated by components, and a set of paired cues to validate the viability of reinforcement learning as a model of behavioral adaptation-- and/or as the result of neural reward pathways.
The researchers made use of electroencephalograms (EGG), separated by components, and a set of paired cues to validate the viability of reinforcement learning as a model of behavioral adaptation-- and/or as the result of neural reward pathways.
They hypothesized that because learning through trial error is inefficient and potentially dangerous for humans, other mechanisms of learning may be highly valuable. The ability to learn from non-experiential exposure was found to be not only advantageous, but also rewarded (assimilated through neural pathways.)
The researchers studied instruction or the lack of it as their model for behavioral adaptation through neural pathways. The participants were given probabilistic assignments, and the events were tracked and scored for later analysis. The test groups were separated as follow:
Test group 1: received feedback indicating response reward--after each task.
Test group 2: received the same feedback given to test group 1 and additional instructions about the reward probabilities, before they conducted another task.
The researchers found that "instruction eliminated participants’ reliance on feedback." By contrast, negative responses related to feedback and the event-related potential, both associated with neural reward prediction errors keep adapting with experience. Both groups benefited from experience.
The research findings suggest that while instruction may influence or control behavior right away, certain neural responses must be learned from experience.
Test group 1: received feedback indicating response reward--after each task.
Test group 2: received the same feedback given to test group 1 and additional instructions about the reward probabilities, before they conducted another task.
The researchers found that "instruction eliminated participants’ reliance on feedback." By contrast, negative responses related to feedback and the event-related potential, both associated with neural reward prediction errors keep adapting with experience. Both groups benefited from experience.
The research findings suggest that while instruction may influence or control behavior right away, certain neural responses must be learned from experience.
No comments:
Post a Comment