Puppy Forum and Dog Forums banner

1 - 20 of 34 Posts

·
Registered
Joined
·
869 Posts
Discussion Starter #1
Circular logic/reasoning is a kind of logical fallacy. Definitions are circular logic. The definition of reinforcement is circular, but lets examine the concept of a reinforcer - something that increases behavior.

The circular logic of a reinforcer goes like this -
How do you know it was reinforcing? Because it increased behavior. Why did it increase behavior? Because it was reinforcing. How do you know it was reinforcing? Because it increased behavior. etc etc

How do you know ahead of time if you are applying a reinforcer? You don't, and you will never know beforehand. You only know afterwards from the result. So what use is it if you can never be sure that what you are applying is going to work as you want?
 

·
Registered
Joined
·
3,722 Posts
It's the end result that determines if you used a reinforcer or a punisher. For example, if you have an aggressive dog and he lunges and barks at another dog and you give a leash correction, it might only wind the dog up more and make him worse. So you intended to punish, but actually ended up reinforcing.

I'm not sure what you mean by "what use is it"? You can know from experience what will be reinforcing and what won't, but no, there is no way to be 100% certain something will be reinforcing. You can also reward a dog without actually reinforcing behaviour.
 

·
Banned
Joined
·
1,765 Posts
It's more like reward and punishment. You reward behavior hoping to reinforce it. You punish behavior hoping to discourage it.

"reinforcement = increased behavior" is more of a definition rather than logical steps.
 

·
Registered
Joined
·
161 Posts
You don't, and you will never know beforehand. You only know afterwards from the result. So what use is it if you can never be sure that what you are applying is going to work as you want?
I guess that we can never be sure if a certain method works or not with some dogs in some situations. All we can do is to gather as much info as we can and think of the chances of success each method has. And then try it, and then by analyzing the results, determine what needed improvement, what works and what doesn't.
 

·
Registered
Joined
·
869 Posts
Discussion Starter #5
If anyone's followed my posts in the past, you'll know that I'm a big OC learning nerd, but I've given it some thought and and I'm struggling with the self-fulfilling nature of OC. In otherwords, it only qualifies as OC if what you did worked. If it didn't work, then it wasn't OC's fault because it didn't qualify as OC.



I guess that we can never be sure if a certain method works or not with some dogs in some situations. All we can do is to gather as much info as we can and think of the chances of success each method has. And then try it, and then by analyzing the results, determine what needed improvement, what works and what doesn't.
But see, we can only determine what has worked in the past, it doesn't tell us about what will work now or in the future. At best the past gives us a point of reference but for example, if a dog habituates to an aversive stimulus, then obviously the punishment becomes less effective. The problem is that's not something you can know beforehand. It only becomes clear when one day you're jerking and yelling and it doesn't happen like it used to.
 

·
Registered
Joined
·
9,391 Posts
Punishment is supposed to decrease OR stop a behavior, so if you continually jerk the leash, then the punishment is not effective. The point that you made - habituation, the need for precise punishment, and the possibility of escalating punishment ... are all excellent reasons for avoiding punishment.

I do not recommend this, so I shouldn't write this, b/c people may try it, but effective punishment is quick, decisive, and intense enough to Stop the behavior. I have seen experienced trainers who have the timing to give decisive corrections which are not brutal and which do not intimidate the dog. It is amazing to see, b/c the dog doesn't cower or submit, just kinda says, "Huh?" and stops. However, I know that I don't have that type of timing or touch, so I advise against punishment.
 

·
Registered
Joined
·
12,761 Posts
So what use is it if you can never be sure that what you are applying is going to work as you want?
I'm not clear what you want to get at here. Are you saying you don't know what the end result should look like? If you're not saying this, well, you always have something to compare your progress too, with objective analysis of the facts, and comparison.
That, to me, is what makes training more a science than an art.
 

·
Registered
Joined
·
4,977 Posts
Circular logic/reasoning is a kind of logical fallacy. Definitions are circular logic. The definition of reinforcement is circular, but lets examine the concept of a reinforcer - something that increases behavior.

The circular logic of a reinforcer goes like this -
How do you know it was reinforcing? Because it increased behavior. Why did it increase behavior? Because it was reinforcing. How do you know it was reinforcing? Because it increased behavior. etc etc
The second question is not a definition. You're asking about the process of why reinforcement works in an dog, not "what is it called when something cause behavior to occur more often?"

Asking for why something works (process) is not the same as asking what something is (definition).

The "why" reinforcement works the way it does likely goes into brain chemistry, states of mind, the observation abilities of the dog, etc. "Because it's reinforcing" does not answer the question of "why it increased behavior". It's just restating the definition of what a reinforcer is (something that increases behavior) in "answer a question" form. The answer to "why" would go into why a dog would repeat the behavior, i.e. why does he think to do it again, and why is it that some dogs, especially those exposed to shaping, might try a different behavior to see if it, too, brings a reward? That's getting more inside of why reinforcement works like it does (and it's inverse, punishment), but it's beyond the reach of what OC is designed for (observing what causes behavior to change and exploiting it).

As far as the past vs future - the future is never known. The past is always certain. Nothing can tell you with absolute certainty what will happen in the future. I believe this especially true with another living, thinking creature in the mix in an environment of "random" events that you have only very limited control over! The past is always certain, because it already occurred. I don't think that's just OC - it seems likes it's just life.
 

·
Registered
Joined
·
9,307 Posts
I have seen experienced trainers who have the timing to give decisive corrections which are not brutal and which do not intimidate the dog.
and the above is aversive stuff that cannot be taught, it comes from dog after dog after dog, action/reaction, action/reaction etc. That's why it's barbaric to try to online explain on how to correct a dog with prong/e-collar type work It's definitely gonna lose something in the explanation.

Trust me on this (hate people that say trust me) I have made correction mistakes training dogs back in the Jurassic days. It's the mistakes that help a trainer learn/hone his trade. Problem is there are trainers that do not have the learning ability.
 

·
Registered
Joined
·
161 Posts
and the above is aversive stuff that cannot be taught, it comes from dog after dog after dog, action/reaction, action/reaction etc. That's why it's barbaric to try to online explain on how to correct a dog with prong/e-collar type work It's definitely gonna lose something in the explanation.

Trust me on this (hate people that say trust me) I have made correction mistakes training dogs back in the Jurassic days. It's the mistakes that help a trainer learn/hone his trade. Problem is there are trainers that do not have the learning ability.
I agree that it's dangerous when explaining how to implement corrections on the internet, although I'm not sure what's best coz some people would still try it by themselves even without explanations, so that I'm not sure if it's better to keep them informed or keep them from learning. But in any cases, I guess that it's better to always tell them to find a trainer if they must really use it, so that they have proper guidance and won't misuse it.
 

·
Registered
Joined
·
10,186 Posts
Speaking as a logic nerd, your problem is you're applying logic to behaviorism. Living creatures don't operate by pure logic.
 

·
Registered
Joined
·
19,018 Posts
I think you are overthinking things. And this is coming from a mathematician who has taken many a class on logic.

Living beings have a history and you can use that history to predict behavior. 100%? Of course not.
 

·
Registered
Joined
·
5,406 Posts
Circular logic/reasoning is a kind of logical fallacy. Definitions are circular logic. The definition of reinforcement is circular, but lets examine the concept of a reinforcer - something that increases behavior.

The circular logic of a reinforcer goes like this -
How do you know it was reinforcing? Because it increased behavior. Why did it increase behavior? Because it was reinforcing. How do you know it was reinforcing? Because it increased behavior. etc etc

How do you know ahead of time if you are applying a reinforcer? You don't, and you will never know beforehand. You only know afterwards from the result. So what use is it if you can never be sure that what you are applying is going to work as you want?
You can have a pretty good idea of what may reinforce your dog. It is useful because it allows me to build and test a theory (what reinforces my dog). Usually it works, but sometimes it requires more thought on my fault if my dog finds something less reinforcing (or more reinforcing) that I predicted.
 

·
Registered
Joined
·
5,406 Posts
If anyone's followed my posts in the past, you'll know that I'm a big OC learning nerd, but I've given it some thought and and I'm struggling with the self-fulfilling nature of OC. In otherwords, it only qualifies as OC if what you did worked. If it didn't work, then it wasn't OC's fault because it didn't qualify as OC.


But see, we can only determine what has worked in the past, it doesn't tell us about what will work now or in the future. At best the past gives us a point of reference but for example, if a dog habituates to an aversive stimulus, then obviously the punishment becomes less effective. The problem is that's not something you can know beforehand. It only becomes clear when one day you're jerking and yelling and it doesn't happen like it used to.
Life is full of "what ifs?" So is dog training. It's not an exact science and neither we nor our dogs are infallible, perfect beings (well, we may be perfect in some philosophies, but we have a hard time finding that perfection.)
 

·
Registered
Joined
·
869 Posts
Discussion Starter #15 (Edited)
That, to me, is what makes training more a science than an art.
As far as the past vs future - the future is never known. The past is always certain. Nothing can tell you with absolute certainty what will happen in the future. I believe this especially true with another living, thinking creature in the mix in an environment of "random" events that you have only very limited control over! The past is always certain, because it already occurred. I don't think that's just OC - it seems likes it's just life.

If something is based on science, then the result should be predictable. If OC is science, then we should be able to predict the result right? Or is my understanding of science incorrect? I guess my occupation is hard science whereas psychology is more of a soft science. When I work with electronics, I know that if I put certain resistor values somewhere, I will get a certain voltage, without fail, everytime. I harness these laws to make something work. In OC, if the only way you know of something working, is from the result afterwards. You don't have that infallible way to harness a predictable result as in the hard science field. Sure, most of us have success because we have a general idea, and a feel for how to do it, but that's not science, that's art.

It's like if I was evaluating whether or not a coach is a good coach. I set my definition for "good coach" as someone who has a career winning % of over 90%. There's no way to know beforehand if the coach is a good hire. You only know after the coach's career. What use is that?
 

·
Banned
Joined
·
1,765 Posts
A lot of people would say psychology isn't a real science. While a lot of psychologists will claim it's a science.

Also back to your first example, if it's cause and effect, you can pretty much always try to trace the cause from the effect and vice versa. Like I can say "How do I know I heated up the water, because it boiled. Why did the water boil, because I heated it up" I don't really see that as circular reasoning.
 

·
Registered
Joined
·
20,765 Posts
It's not quite the same thing. You know you heated up the water because you turned the stove on and put a pot of water on the burner, and that always heats up water. I find the whole thing pretty confusing too.
 

·
Banned
Joined
·
1,765 Posts
You know you heated up the water because you turned the stove on and put a pot of water on the burner, and that always heats up water.
that's kind of the point......even if it's something you know for sure, you can still trace the effect to the cause.

The issue with circular reasoning is more just that it can't be used to prove anything.
 

·
Registered
Joined
·
19,018 Posts
If something is based on science, then the result should be predictable. If OC is science, then we should be able to predict the result right? Or is my understanding of science incorrect? I guess my occupation is hard science whereas psychology is more of a soft science. When I work with electronics, I know that if I put certain resistor values somewhere, I will get a certain voltage, without fail, everytime. I harness these laws to make something work. In OC, if the only way you know of something working, is from the result afterwards. You don't have that infallible way to harness a predictable result as in the hard science field. Sure, most of us have success because we have a general idea, and a feel for how to do it, but that's not science, that's art.

It's like if I was evaluating whether or not a coach is a good coach. I set my definition for "good coach" as someone who has a career winning % of over 90%. There's no way to know beforehand if the coach is a good hire. You only know after the coach's career. What use is that?
You do know 100% of the time that a reinforcer will increase the likelihood of the behavior happening again. If you're asking will this treat I am giving my dog act as a reinforcer 100% of the time then you can't know 100% certain. All I can say is in the past giving a treat has been reinforcement and therefore I expect it will in the future.

I will be honest in that I don't understand the point of trying to pin training down 100% into exact truths. Any time you throw in a living being (not to mention two beings of two different species) you will never get something 100% of the time.

Then again I am a mathematician AND an artists. I don't think the world can operate without both.
 

·
Registered
Joined
·
4,977 Posts
If something is based on science, then the result should be predictable. If OC is science, then we should be able to predict the result right? Or is my understanding of science incorrect? I guess my occupation is hard science whereas psychology is more of a soft science. When I work with electronics, I know that if I put certain resistor values somewhere, I will get a certain voltage, without fail, everytime. I harness these laws to make something work. In OC, if the only way you know of something working, is from the result afterwards. You don't have that infallible way to harness a predictable result as in the hard science field. Sure, most of us have success because we have a general idea, and a feel for how to do it, but that's not science, that's art.
Science is not 100% able to predict the future. Let me put it this way, meteorology is a science. Is it capable of predicting with 100% accuracy tomorrow's weather?

Why is that the case when what you say is also true? Because we don't have all the known information, all the known variables and can't predict how those unknowns will interact with what we do know, and why when sometimes all the ingredients ARE present, the expected result STILL doesn't happen.

Does that mean meteorology is not a science?

Another example, math is a science. However, a whole subset of math is about anything BUT absolutes - i.e. probability theory. Does that mean math is not a science. Does it mean probability is useless and has no real value? Does it mean we should not try to quantify it?

Electricity isn't exactly 100% predictable either, otherwise, we wouldn't have power surges, spikes, could predict exactly when and where lightning will strike, etc. Heck, don't some atoms have variances (isotopes)? I wonder if the physical world is as absolute as we humans want to make it out to be.

To me, the same is true with operent conditioning. We don't know all of the variables with absolute certainty. OC says, basically, to throw them out. That's not satisfactory to me because, what if I could use some of them to be more effective? However, that is a possible reason why it is not 100%, why not all dogs have the same motivators and punishers, why some techniques work better with some dogs. Wally could have an upset stomach one day. Does that mean OC doesn't work because he doesn't half kill himself trying to get a piece of chicken? No, it means there's other variables than those four boxes. That doesn't make the four boxes any less invalid.

As far as your example with a coach - that's because you're using a trailing indicator. You're looking at something that already happened. However, if I know a coach runs his offense with, say, a vertical passing game and I know I have a QB that throws like a girl and precise, but not fast, WR - I can predict he'd have a harder time than a coach that runs a team like Bill Walsh and his short precision passing attack.

I don't know with absolute certainty, but I can improve my odds, so to speak. Sometimes in life, I think you just have to accept that as the most control you can get at the moment and adjust as you go. Adaptability is supposed to be nature's greatest gift, no?
 
1 - 20 of 34 Posts
Top