This is my response to Ulop and Roy on our Community Network on Connectivism:
Great that we have come to some common themes on learning.
I think it’s imperative to distinguish human learning from the learning “that may reside with non-human appliances”.
An example is when you go to ATM to get money. The ATM has “learnt” how to issue the correct amount of money you have keyed in and issue you with the money with receipt. So the ATM processes your order based on an algorithm and a process run by the machine and computer, and that is taught by human. Such processing of money for you is similar to concepts adopted by artificial intelligence. Now, what happens if someone put fake money into the ATM. So you will also receive those fake money when you use those ATM. For the ATM, it has “learnt” to give you the money you have requested, but it hasn’t learnt to check if the money stored was fake or not. So, if you receive the fake money, it isn’t the “fault of the ATM” and the ATM is 100% accurate in learning and “highly intelligent” in accordance to our human initial design.
Once you realise such problem of fake money, you may then re-design the ATM so that it could check for fake money. So, you would then teach the ATM to check the fake money (bank notes) before it issues any money. Through this process of learning by you and your re-design of the ATM, the ATM is learning through you as human how to ensure that only real money is issued to the customer. The ATM itself can only learn how to do the job through human intervention. By itself, it isn’t as smart as human. And sometimes, the ATM may fail to check whether the bank notes are fake or not if there are changes in the design of the bank notes or there could be mistakes made due to the malfunction of the “ultra-violet” detector (say due to failure of the detector) of the ATM. So, you may then rely back on human to check if the money is fake or not.
The above metaphor is again trying to illustrate how smart human are as compared to machine. Similarly, I don’t think there has been any machine that is built which could simulate our digestive system so far, as we could cleanse any toxins through our body organs and eject waste which are useless for us. I would like to learn if such a machine exists in this world which could do all these!
This ATM example illustrates that:
1. Human learns through a biological and a neuro process with the brain (just like the digestion metaphor), and it is different from machine learning in that the machine can ONLY learn when the human teaches it (even if it’s artificial intelligence). You may claim that a machine can do some “learning” by itself, but as the above example illustrates, it must start from human. And a machine may fail to learn if the human doesn’t teach it to learn properly – fake money will be issued to customers without notice or warning, though the ATM is functioning 100% effectively and efficiently.
2. In human learning, there are some common learning principles with non-human learning (animals or even appliances). These include the observable – the Stimulus-response classical conditioning by Pavlov. Classical conditioning is the study of learning which involves reflex responses, in which a neutral stimulus comes to elicit an existing reflex response. Please note that Pavlov’s work on the physiology of digestion, begun in 1879, earned him the Nobel Prize in 1905. He first became aware of reflexes by reading Sechenov’s work while still at seminary, but his own research on what became known as classical conditioning did not begin until about 1902. At this time, while still studying in digestion in dogs, he noticed what he called ‘psychic salivation’ – a dog would salivate before it was actually given food. Since Pavlov believed that digestion involved series of reflexes, he set out to determine what controlled this anticipatory response. Ultimately, his work on conditioning overshadowed the research which had earned him the Nobel Prize.
3. I try to distinguish the human from non-human learning to avoid the confusion arising out of the studying of the non-human appliances, ants, spiders, pests in their life cycle, its ecology from human, especially when we are referring specifically to learning over the digital ecology, the net, virtual networks, and communities. There may be a lot of learning embedded in such social and ecological studies, and so I will leave it to the Biologists, Sociologists and Social Scientists or YOU to investigate. Sometimes, there might be a similar pathway in adopting the “behavioural” approach by observing the behavior of those creatures and generalising them on human. Would this be what Pavlov had tried to do? However, I do think we have overlooked his work on the physiology of digestion. I have now used digestion as a metaphor on learning. I must admit that I don’t know all his work on signal conditioning (and have forgotten what I have read years ago) until you asked me now. Please see Approaches to Psychology by William E. Glassman 2000 (that I bought more than 8 years ago).
Ulop and Roy, I am interested in learning how these could be further explored. I think it could lead to a great concept map which deploy all the learning components as cited by Roy and your critical analysis of learning. Let’s continue…