Neuroscience
Learning action responses
The “brain” — a collection of 25,000 living neurons, or nerve cells, taken from a rat’s brain and cultured inside a glass dish ...
When DeMarse first puts the neurons in the dish, they look like little more than grains of sand sprinkled in water. However, individual neurons soon begin to extend microscopic lines toward each other, making connections that represent neural processes. “You see one extend a process, pull it back, extend it out – and it may do that a couple of times, just sampling who’s next to it, until over time the connectivity starts to establish itself,” he said. “(The brain is) getting its network to the point where it’s a live computation device.”
“Initially when we hook up this brain to a flight simulator, it doesn’t know how to control the aircraft,” DeMarse said. “So you hook it up and the aircraft simply drifts randomly. And as the data comes in, it slowly modifies the (neural) network so over time, the network gradually learns to fly the aircraft.”
personal notes: This experiment could provide key insight into how a "brain" (a network of biological neurons) may learn what actions to send without the need for reward. The input/prediction/output feedback loop would try to minimize prediction error, as is the intrinsic function of neurons and their synaptic processes per Hebbian learning and Spike-Time-Dependent-Plasticity. As output (i.e. actions) have an effect on the environment, the proper action selection choice may help minimize future prediction errors. In essence, this relates to proprioception, having a model of where one is in the world and updating this model based on actions to better predict future stimuli.
Connectome of learning centre
helper: to read this paper, learn these acronyms by heart:
- MB - Mushroom Body
- KC - Kenyon Cell
- PN - Projection Neuron
- DAN - DopAminergic modulatory Neuron
- OAN - OctopAminergic modulatory Neuron
- MBIN - Mushroom Body Input Neuron
- MBON - Mushroom Body Output Neuron
- APL - Anterior Paired Laternal neuron
The fact that single-claw KCs appear earliest in development suggests that a top priority, initially, is to assure that a complete set of signals is relayed to the MBONs, which is not guaranteed with random wiring.
personal notes: Pre-wiring exist (beyond chance) to ensure that output neurons receive a complete set of input. Relate this to the study on innate knowledge in newly hatched birds that categorize the shadow cast by the same model bird as either predator or parent depending on its direction of movement (when flying in one direction, the model bird had wings in front, indicative of predator; when flying in the other direction, the model bird had wings in back, indicative of parent). Note, experiment replication efforts of the so called hawk/goose effect seem to have failed. As such, please take these mental notes with a healthy measure of scepticism.