By combining all previous lectures, we will today see how and why eventually our brain learns and changes. So neurons are also considered to be plastic & changing devices. The plasticity of the brain makes us learn, this is what makes us actually learning machines. So, what exactly changes in your brain while you learn?... (click on Read More to continue).
One of the major purposes of learning is to generate useful predictions. Then, by heavy learning one doesn't just excel in a specific field but he/she can also predict his/hers next move or outcome or consequence of his/hers action or choices. Secondly, we use learning to categorize the world into homogenous sets of "things". Finally, we use learning to create "consenus" among us, mainly by language, for successful interaction.
Learning as an active process
The twin cats example tries to exemplify the notion above such as that even sight (or probably other main senses as well) are heavily dependent on learning. In this case, it was proven that movement is essential for perceptual learning.
The twin cats example tries to exemplify the notion above such as that even sight (or probably other main senses as well) are heavily dependent on learning. In this case, it was proven that movement is essential for perceptual learning.
To learn to see, you have to act. This is called the action-perception loop (first you act, then you perceive, once you perceive you act better and so on (=loop)). Above, we have two new-borne, identical twin kittens. The one has its legs free-hanging out of the basket and can walk. The other is passively moved by the activity of the first cat (check motor mechanism above them). The amazing outcome is that once you free these kittens (after 3 weeks) and let them freely walk in an unobstructed environment, in an empty room, the "active" one can walk and do things just normally while the other one keeps bumping onto the walls. It can't see! It behaves completely like a blind cat. And this is shocking since both cats are genetically identical. The fact that the "passive" cat was not moving physically, somehow affected its ability to interpret the visual world, in other words, the natural movement was correlated to the visual system and enabled the "active" one to develop a "correct" perception of the world (if we agree of course that there is indeed such a thing as "correctness of perception" or even... "the external world"...!). So eventually, in the case of blind kitten, movement and visual input did not correlate in order to form a coherent image of the world. When vision and action go together, you learn to see. So seeing is not a passive thing, we're not a camera (!!!), we're an active machine that learns to interpret the world and this interpretation only becomes correct when you move (at an early age). Movement is the most primal and primitive aspect of behavior.
Reality
In order to develop concepts of reality (again, if such a thing actually exists out there) one uses very little information and when combined with stored information in memory as well as learning, then "some reality" is being built.
In order to develop concepts of reality (again, if such a thing actually exists out there) one uses very little information and when combined with stored information in memory as well as learning, then "some reality" is being built.
In any case, reality is surely an agreement between humans. Let's see the problem of hearing impairment. Some people with hearing problems just listen to noises, to sound "waveforms" of no interpretable meaning. So cochlear implants are planted inside the ear to restore exactly this process of interpretation of the physical "waveform" of sound to a meaningful signal. So the brain reconstructs various kind of information in order to produce "understanding". And understanding stems from our need to learn something by first categorizing it into a homogenous existing category of "things" which facilitate the learning process and prevent our memory from storing redundant or duplicate information (sorry, but I can't help it but inserting some personal comments here and there since I've had quite a few readings on Concept Formation in the past, sorry if you find this annoying, I don't want (I can't!) substitute Mr. Segev, I just try to add/explain things a bit more).
How do the anatomical characteristics of the brain support learning and plasticity?
Once again, the big father of anatomy, Santiago Ramon Y Cajal, cannot be avoided!
Once again, the big father of anatomy, Santiago Ramon Y Cajal, cannot be avoided!
Of course he was wrong about nerve cells multiplication/reproduction but in general it's clear that he was clearly interested in the underlying mechanisms of brain/learning. He said that there must be some change in the brain when learning. And this is called structural plasticity, the underlying structural change, below learning and memory. Now, let's discuss some possible underlying mechanisms.
Functional Plasticity
It's the strenght or weaking of existing synaptic connections. The Hebb Hypothesis (1949):
It's the strenght or weaking of existing synaptic connections. The Hebb Hypothesis (1949):
So if cell A is involved in firing cell B continuously (so both cells are active) then somehow cell A becomes more effective in firing spikes to cell B (my note: well, isn't it the same principle here, just at a mini-/micro- level of understanding? I mean, again it's all about movement and action. Active cells progress and prosper. Inactive cells decay. No? Very very interesting and mind-blowing implications for actual macro-like (so to speak) human actions and behavior...). Of course it's an hypothesis but a very informative one that talks about causality at such a low-level of anatomy. My fact of firing electical spikes to cell B and its "attitude" of being active and accept this communication, somehow strengthens the in-between relationship, causing myself (cell A) to act more efficiently. Just amazing...!
So (again), let's say that the axon of a cell A (pre-synaptic cell) fires an action potential (spike, for the new-comers). Cell B (the post-synaptic cell) responds it withn an EPSP. Let's also say that this happens "before learning", that this is the basic efficacy of the synapse. So a strong 100mV digital spike (remeber, all-or-none, 0/1) generates a "humble" and attenuated 2mV EPSP. Hebb says that if this is going on for a sufficient long time, something changes and the connection becomes more effective.
So after learning, Hebb says, cell A increases the probability of cell B to also fire but this time, a bigger EPSP for the same spike to make a stonger synapse, to create a bigger depolarization. So what could be the mechanism for such a thing? We know today that one possible mechanism is that of insertion of additional receptors to the post-synaptic membrane. We know today that due to plasticity of its synapse, the same synapse may now have post-synaptically more receptors
But there could also be true the fact that more neurotransmitters are transmitted by the pre-synaptic cell. In any case, with new techniques applied today, one can in-vivo see where do synapses change and how this alteration takes place.
Spike Timing-Dependent Synaptic Plasticity (STDP)
This particular condition makes the synapse stronger. It says that you first activate a spike, in the pre-synaptic cell (with the electrode) - blue electrode). Then, with the red one, you activate a EPSP at the post-synaptic cell. Suddenly, after enough repetition, the synapse becomes stronger. This is called Long-Term Potentiation (LTP), the synapse is potentiated and becomes stronger in time (minutes/hours/days/lifetime...). If you do it in the reverse order of time, meaning that if you first activate the post-synaptic cell and then the pre-synaptic one, the synapse grows weaker and we have Long-Term Depression (LTD).
This particular condition makes the synapse stronger. It says that you first activate a spike, in the pre-synaptic cell (with the electrode) - blue electrode). Then, with the red one, you activate a EPSP at the post-synaptic cell. Suddenly, after enough repetition, the synapse becomes stronger. This is called Long-Term Potentiation (LTP), the synapse is potentiated and becomes stronger in time (minutes/hours/days/lifetime...). If you do it in the reverse order of time, meaning that if you first activate the post-synaptic cell and then the pre-synaptic one, the synapse grows weaker and we have Long-Term Depression (LTD).
For these mechanisms to work (LTP and LTD), there is a very narrow time window of +40ms and -40ms, respectively. If you wait more, say one minute, nothing happens. This theory though can't explain other things, like the Pavlov case learning which is a rather longer-term mechanism of plasticity (count in seconds between experimenter's ringing a bell and dog's saliva to start spilling).
Today, exactly because we know these mechanisms experimentally, we can write the equations that describe LTP and LTD in order to make models.
Structural plasticity
We refer to structural plasticity as the anatomical/morphological brain tissue changes that are correlated to learning. In 1967, Globus & Scheibel found that sensory experience affects dendritic spine variation (creation or deterioration) in rabbits.
We refer to structural plasticity as the anatomical/morphological brain tissue changes that are correlated to learning. In 1967, Globus & Scheibel found that sensory experience affects dendritic spine variation (creation or deterioration) in rabbits.
They saw correlation between structure and function, or the capability to improve behavior. But these techniques came from fixed brain dead tissues. Today miracles can happen, i.e. with the 2-photons microscope method, which allows to view in a very fine resolution a very specific part of a living brain. By this method we learned that dendritic spines are born (appear) and die (disappear, for more (semi-transient spines) or less). Spines trying to find a connection and succeed in doing so, the survive and prosper. Others, not so lucky or potent, they dissolve. (yes I know Darwin is smiling right now, this is pure evolution in the foundations of our anatomy). So there is a constant change in the brain. Does it relate to learning? Definetely, yes!
So we have today a direct proof, just by looking in-vivo on the behaving brain using a 2-photon microscope, that new structures grow or disappear all the time. New spines are "born" constantly. During learning tasks or enhanced environment more spines are created. These new spines are associated with new synapses and thus create new functional networks (new memories?) that code information. Here it is important to be said that this coding is quite unique and "personal" for every brain. Not every brain will produce the same spikes and synapses for the same stimulus. So we all code reality in a different way by default and then we come to an agreement (mainly a linguistic one) as of to what "red" or "tree" or "reality" mean in order to be able to develop societies.
Neurogenesis
Do new-born cells exist in the mammalian adult brains? This is called neurogenesis and the answer is yes, contrary to the common belief from the past (we can use cell-staining methods to track them).
Do new-born cells exist in the mammalian adult brains? This is called neurogenesis and the answer is yes, contrary to the common belief from the past (we can use cell-staining methods to track them).
It was also shown that the most challening task you face, the more new cells your brain creates.
So new cells are born as stem cells, they enter the neural network and they are either stable and integrate to it by "maturing" (actually, by trying to help our brain solve complex problems) or they are unstable (not completely understood yet) and vanish in oblvion... In any case, we certainly need to better understand neurogenesis for medical reasons (Parkinsons, Alzheimers or rehabilitation of strokes and brain injuries).
To summarize, at least one thing is clear. The old notion of "If you don't use it, you lose it", was 100% correct. The more you use the network, a challenging environment, an intellectual task, motor activity, music (play and hear), etc, the more new connections you have (spines), the more new cells you give birth to (in the hippocampus). So the interaction between activity behavior and this plastic system that likes to change so much is about making new spines or cells and synapses, making them stronger and weaker. So the message should be: BE ACTIVE! (:-))...