Joined: Oct. 2005
|Quote (GaryGaulin @ June 10 2018,19:13)|
|Quote (N.Wells @ June 10 2018,08:04)|
|Making a guess is not a sign of intelligence, and it is certainly not a requirement of intelligence, because random responses do not require forethought or planning.|
If there was no "forethought or planning" happening then the system would not know when it needs to take a guess.
The ID Lab 6.1 model shown in a video has an internal view of the external world all mapped out "in its mind" and intuitively knows which direction it should be heading, etc.. You conveniently ignored all that, and my getting an important signal ratio matching literature for live rats planning how to get the food without getting a shock.
|Quote (N.Wells @ June 10 2018,08:04)|
|However, learning from the results of guesses (learning from trial and error) is one of the features of intelligence.|
Learning from the results of guesses is what happens when all four of the circuit requirements are met. This might be helpful to add to the text, but you talk like "learning from mistakes" and such are not accounted for.
Your attempts to discredit the model are very annoying. If it were not for their occasionally being useful then I would have no reason to bother with them.
The exercise at least shows how to "defend" a "scientific theory" and how having a model behind it (diagram of a circuit and coded program) makes all the rest a semantics argument over how to word an explanation for how it works that instead uses sentences.
I now need to make clear sense to those who are up against a brick wall of sorts caused by the data stream through the brain still leading to motors that produce vocal sounds while automatically using body motions to match what's being expressed in their thoughts, not digital text output. What is needed is an easy to conceptualize cerebellum that works with existing cortical models.
Adding guess to the addressing side of the RAM requires additional code that will slow down the program a little but doing so allows up to a million 10 bit wide addresses, where each is an 8 bit location containing (100% guess derived) motor action data. The newest model I did not yet make a video for and has no cerebellum is still able to get around on the spatial navigation network of it cortex alone, but has the poorly coordinated features of human patients with serious cerebellum damage.
I without knowing it (except for guessable address inputs) have been experimenting with what happens with and without a cerebellum. I previously stuck not knowing how to explain what it was I was experimenting with, now it's easy!
This is a breakthrough I was hoping for in regards to being clear. And the necessary reference information is a relatively easy to make sense of YouTube video. The way "science" gets around these days is truly amazing. I love it.
Hogwash. Many motile bacteria utilize random walks in a way that is the equivalent of making a guess. They can then use the current strength of a chemical signal vs the previous strength of the signal to determine which direction to head in. It may look like intelligence in operation, but none of that is cognition and reasoning - all of it is biochemical reactions that result in chemotaxis. Specifically, the bacteria are too small to detect gradients by comparing concentrations at their head end to their tail end. Instead they do brief random walks or tumbles in the middle of straight runs to obtain two readings some distance apart. The onset of tumbling motion is chemically controlled:
|the direction of flagellar spin is regulated in part by methyl-accepting chemotaxis proteins (MCPs). The extracellular domain of an MCP is a chemosensor, which responds to changing concentrations of its target molecule by a shape change. The signal embodied by that shape change is transduced first through a membrane-spanning domain, then to one or more small linkers called HAMPs, and finally to a kinase control module, which can turn on, or turn off, a kinase, which then either does, or doesn’t, phosphorylate another protein. The ratio of the phosphorylated to unphosphorylated protein is the final determinant of the direction of flagellar rotation, and hence the movement of the bacterium.|
You and I have no guarantee that your model is grounded in reality: you merely think it ought to work that way. Your language about "4 requirements of intelligence" is really confused and ad hoc (e.g. your language about "motors to control"). You do not talk about learning from trial and error, you just say that one of the "requirements" (which is not what you mean) of intelligence is "the ability to take a guess", which is not true. Many non-intelligent systems do the equivalent of "making a guess" (e.g. Neato vacuum cleaners, pollen blowing in the wind, larvae floating in the ocean), but learning from trials and errors is far more a hallmark of intelligence.
As annoying as my comments are, they are entirely deserved, because you have yet to demonstrate that you have anything of worth.
Your labelling something in a simulation with the name of a brain part does not make it that brain part nor does it guarantee that it is modelling that brain part accurately.
|The newest model I did not yet make a video for and has no cerebellum is still able to get around on the spatial navigation network of it cortex alone, but has the poorly coordinated features of human patients with serious cerebellum damage. |
SnowTotal = SnowTotal + Snowflake
Loop until SnowTotal >= Blizzard
Hey - I just modelled a blizzard!