The 1 to 2 second rule: Sucker Error and Scientific Authorities

stopwatchA common sucker-error of modern training and behavior starts at the lowest level of understanding. Here’s a question that any trainer should be able to answer concisely.

From the moment a behavior occurs, how long do I have to present a tangible reward or punishment? You may have heard all kinds of statements about this and here are some examples that are common among modern “scientific” dog training and behavior experts. These are examples of smart, highly educated people who are anything but experts on this topic. That’s because they are exactly wrong in their statements. No, it’s not me saying they are wrong just to criticize them. It is reality that exposes their error. At the end of the article you, too, will know they are wrong based on your own experience.

  • A punishment must follow within three seconds to effectively inhibit the immediately preceding behavior. Ideally, both crime and punishment should occur in less than a second. The dog must know precisely why it is being punished. The relevance of punishment depends very much on its immediacy. If the dog’s ‘crime’ is immediately followed by punishment, the dog will get the picture. – Ian Dunbar PhD.

Here is a quote from the American Veterinary Society of Animal Behavior (AVSAB) – a collection of veterinarian behaviorists from their Position Statement on Punishment.

  • AVSAB recognizes that both positive reinforcement and punishment require significant skill, effort, and awareness on the owner’s part. Both must be applied as the animal is performing the target behavior or within one second of the behavior to be most effective.

     

    This is from a website for 4 Paws University

  • In order for any of these approaches to be effective, the consequence must be delivered immediately (no more than 2 seconds), and consistently. Where most dog owners go wrong is delivering the reward or the punishment too late or inconsistently (ie, letting the dog pull on leash into the dog park, but punishing the dog for pulling on walks). – 4 Paws University

     

    VCA Veterinary Hospitals information page. Credited to Debra Horwitz, DVM, DACVB & Gary Landsberg, DVM, DACVB, DECAWBM

  • Positive punishment (the application of unpleasant stimulus) is applied to decrease a behavior and not to discipline the pet. It must be administered while the behavior occurs, and ideally just as it starts.

     

    Prepared for the San Francisco SPCA by Ian Dunbar and Gwen Bohnenkamp

  • A dog does not understand the meaning of delayed punishments. If more than 2 seconds have passed the dog cannot connect the punishment with the event.

     

    Stanley Coren paraphrasing a study by Richard L Solomon.

  • “For the group where the punishment occurred directly before the dogs had the opportunity to complete the action of eating the meat, the taboo worked, and the dogs refrained from eating them meat during 30 days of testing. Contrast this to the group of dogs whose punishment was delayed until only 15 seconds after the act. They ignored the taboo and ate the meat after only two days of testing.”

It is obvious that these quotes point to two things – an ignorance of how animals make associations in the real world and the foundational literature of the science of behavior. A common sense, broad view shows that the topic is not actually the amount of time you have to apply punishment before the animal will not associate it with the unacceptable behavior. The broader topic is latency. Does time play a critical factor in making associations? Nobel Laureate Dr. Ivan Pavlov studied this scrupulously. He says no – based on objective observations of reality AND his very precise scientific experimentation.

For starters, most people don’t realize that the same rules apply to positive experiences as to negative ones. AVSAB alludes to that with their 1-2 second rule but apparently has no knowledge of marine mammal training or clicker training. Keller Breland, the protégé of B. F. Skinner, created marine mammal training and called a signal connected to a consequence a “bridging stimulus” because it bridges the temporal gap between the instant the behavior occurs and the delay leading to tangible reinforcement. Breland didn’t understand aversive control and being from the Skinnerian world he never underGoodScienceQuotestood that this latency works for both reinforcement and punishment. The reason I know this better than both Breland and those learned scientists is that my knowledge comes primarily from studying Pavlov’s work enough to know the rules and I looked at the world around me. There are very few behavioral scientists who could teach a dog to do anything reliably on command – and they didn’t learn how to do that during their academic career if they can. Here is a huge point – good science does not propose rules that are not confirmed by objective observation of nature. Good science reveals nature as it is. If reality contradicts science, then it’s not really science.

Getting to the point of the question:
So here’s the answer to our question – according to valid scientific research, how long can an animal go between recognizing a known, paired signal – like hearing the word “NO” – and a tangibly punishing event? (Or a signal and a tangibly reinforcing event) According to these experts we are somewhere prior to 15 seconds but preferably between one and five with several being completely precise in a specific one or two seconds. According to Pavlov the number is almost meaningless. He routinely pushed it out to 30 minutes. Virtually none of his experiments obeyed the 1-15 second rule. No, that does not mean you would punish a dog when you come home if the misbehavior occurred within 30 minutes prior to your return – but it doesn’t mean you can’t if you understand the rules. I’ll discuss that later. Here’s an odd thing – I just asked you to wait until later to discuss a topic. Are you limited by the 1-15 second rule or can you hold your thoughts for awhile and still make a connection? I’ll bet your attention can go a lot longer than that. Dogs can do that, too.

However:
The other thing the scientific people don’t realize is that even if we remove Pavlov the answer to the question is as close as your own front door. Virtually all dogs that live in houses react to knocks on the door or doorbells. There is almost never a time when the sound of the bell or knock presents the guest within the 1-15 second rule. If the experts are right, how can this be? The answer is simple – the experts are dopes. They use a cherry-picked literature and never look outside their catechism. Meaning they do not look at reality to form their opinions. They look at peer-reviewed research. The inherent flaw in that practice is that if the peers are all ignorant they cannot come up with the correct solution.

EG: In the 1700’s there was a prize offered for discovering an accurate way to determine longitude, the east-west distance from any point on the globe. Isaac Newton unequivocally stated that it had to be determined by observations of the moon and would never be discovered with a chronograph. John Harrison created the first maritime chronograph that was so accurate that one could plot longitude with accuracy. Newton was a learned scientist. Harrison was a carpenter by trade. Newton and other scientists opposed giving Harrison the money. Since that day, all ships have used a chronograph to determine longitude. The smartest man in the world, Isaac Newton, was blind-sided by intellectual, academic myopia. He couldn’t see the forest for the trees. Once convinced that longitude had to be determined by celestial observation he couldn’t back down from his pronouncements.

The point of this example is that scientists are not science. They are people who are capable of seeing a single tree and not understanding the forest around them. To be knowledgeable about life and reality the origin of the information is irrelevant if it matches a single criterion – veritas. That is Latin for “truth for truth’s sake.” The problem with modern investigations into behavior is that it lives in a very restrictive world that ignores anything outside their approved literature or observations. In essence, they are in a Skinner Box of their own creation. Too bad for them and those who blindly follow their beliefs.

The Practical Trumps the Theoretical:
So, here’s what you need to know to understand this topic. It isn’t about the timing of the arrival of the reinforcement or punishment. It’s about the animal connecting an event to a consequence based on a number of criteria and timing is only a lesser criterion. Big cats will intentionally freeze to make an ambush more likely to succeed. That is an example of delayed reward. The sight of the prey animal triggers the association between an event and a consequence but does not govern the cat’s immediate behavior. Cats that blundered into always charging at the first sight, scent or sound of a prey animal would be less successful than those that could integrate delayed reward into their hunting schemes.  A coyote may sit waiting at a prairie dog hole all morning because he caught one there yesterday. That implies a 24 latency. Dr. Peter Killeen proposed that an event that happens at 8:00 every morning is closer in your mind than events that happened even a few seconds later – like missing your train. Pavlov demonstrated 24 hour latencies in dogs and you can too – just start feeding your dog at precisely 8:22 every morning and watch the dog’s behavior adapt. It will start hovering around the food bowl at particular times while ignoring the food bowl at all other times. That is nature. It is open to all for observation. It contradicts the concept that there is a time limit on associations.

The difference between the “scientific” perspective and objective observation of nature is two-fold. First, the function of a learned signal that brings more to the table than a single bit of information. In this example, the image of the buck means more to the cat than closing his jaws on a piece of meat. Second, the predictability of an imagined outcome is often far more important to the animal than a knee-jerk response. If you write me a check and I know you pay your debts it’s as good as gold even if I don’t cash it for a week. My behavior changes based on when you gave me the check rather than when I went to the bank…and so it works for dogs and training. It is not the presentation of the treat that identifies the behavior, it is the timing of the click. It is not the bonk with a rolled up towel that identifies the behavior being punished, it is the timing of the word “NO”. If you do not know that fact you will never be able to skillfully control behavior in the real world. To be even more outrageous, being sloppy about the presentation of the actual consequence is a good training strategy if you are trying to break up a long-standing unacceptable behavior. Being sloppy about presenting the “marker” (or not being aware of its importance) is a good way to fail, across the boards..

Final Answer:
The final answer to this overall question is going to be answered by you based on your personal knowledge of behavior. It’s not about me or some other expert telling you a rule that you can swallow. It’s about objective observation of an everyday event that will provide the true answer instantly. Tell me the amount of time it takes between the ringing of a doorbell and when you actually open the door. Unless you are expecting company or it’s Halloween, that latency is almost never less than 30 seconds. (To prove this, consider the surprise you experience when the door is opened as your finger is about to press the bell button. You are surprised  because there is almost always a delay before someone can get to the door. ) Despite that whopping big number, virtually all dogs that live in a doorbell equipped house learn the association.  Once you pick your number you’ll know what Pavlov knew. It’s not about the delay between recognition of a learned signal and the predictable event. It’s about establishing what criteria lead to strong associations. If you are starting to guess that it’s a great deal more than 1-2 seconds you’ve stepped off into the practical world of training.

 

 


15 thoughts on “The 1 to 2 second rule: Sucker Error and Scientific Authorities

  1. When learning punishment or reward maybe delayed AS LONG AS NOTHING HAPPENS (impossible in real life)BETWEEN THE REQUIRED BEHAVIOR AND THE SUBSEQUENT CONSEQUENCE. This is another reason to bridge possibly using a whistle that can be blown from the time the behavior happens to delivery of reward.Pavlov and classical conditioning has to do with involontay responses. Operant conditioning and Skinner have well documented the effects of delayed reinforcement.

  2. Carlo,
    The point of this post is that latency is relatively unimportant in nature. Predictability of outcome is far more important. As for your statement about operant vs. respondent conditioning, that dichotomy does not exist in nature. It is a human construct. Skinner never studied anything meaningful about latency because he never used it in the real world. Meaning documentation is not a means of getting valid information. That is why I cited people who have academic knowledge, yet offer false information. Who do you think they got it from?

    • Gary,there is nothing on this earth that explains any phenomena better than science.This does not make a scientist a dog trainer but it sure hepls the trainer to understand what he is talking about. I read and shared your feelings about Karen Pryor and clicker training. but I can not follow you on this one.Skinner was the begining some 60 years have gone past since and operant conditioning has been verified on “normal” settings outside the lab showing pros and cons.I would not be so sure that classical conditioning or operant conditioning don’t count much or at all “in nature” (in nature we are talking about Canis Lupus Familiaris and village people).Food avversion is just one example of classical conditioning the only responses that can be elicited out of a classical conditioning paradigm are ones that rely on responses that are naturally made by the animal! A dog learns to hunt following his instinct shaped by experience that’s operant!

  3. Carlos,
    Your objections are confusing. First, behavioral science is not actually science. So I can agree with your statement but prove that Skinner wasn’t a real scientist. It does not help a dog trainer to understand Skinner, et al. They lied. They created an agenda driven ideology that is easy to reveal. It’s in the literature. They can’t remove it once it’s in peer-reviewed publications. I can give you the paper trail if you wish.

    Now on to this topic – you have ignored the point of the post. Latency is not an issue – yet this specious one-to-two-second rule is preached by behavioral scientists. No operant can form unless there is a respondent association acting to connect events to consequences. The classifications are human constructs that imply a division where none exists. It’s one organism. That an instinctive reaction might be influenced by external events does not automatically make the behavior an ‘operant’ as there are components within the behavior that are not subject to reinforcement or punishment. An operant is a behavior defined by its consequences. If consequences do not modify the behavior, it’s not an operant behavior – which shows the folly of creating definitions that paint oneself into a corner.

    • Gary, the time interval is very important when learning a new behavior, if anything because in that time interval, anything may happen and be reinforced by the reinforcer delivered too late. So the longer the latency the greater the chances that something occuring during the time delay be reinforced in place of the intended behavior.

      • Carlo, I would suggest that you are too heavily invested in the catechism of ‘behavioral science’ yet do not have the nomenclature mastered. What ‘time interval” are you talking about? My original post is about latency between an event and a consequence and the generally mistaken conclusion about it. You are now talking about the process where a creature discriminates and pairs environmental signals to specific consequences. What you are missing is that the process of discrimination eliminates not contingent stimuli over a series of repetitions. You are attempting to argue that hudreds of millions of dogs, generation after generation, can’t associate a doorbell with someone arriving at the door. That’s a difficult argument to support as it is so obviously incorrect.

    • Carlo, you still don’t get it. The latency between a discriminative stimulus and the arrival of a tangible event is irrelevant. The contiguity of the two is the most important criterion. It does not depend on how much time passes. Try this experiment. Feed your dog at exactly 5:00 each day for two weeks. Then watch your dog’s behavior start to change relative to 5:00. If a dog can sense aw 24 hour cycle, why would any lesser amount of time be a problem?

  4. Makes total sense! My dog, Monday – Monday, wakes up at 6:50am ready to eat and then go back to bed. Thank you for sharing your knowledge!

  5. My alarm used to go off at 6 every morning until my cattle dog figured it out and at 5:50 every morning there would be a nose sniffing my ear to let me know what time it was. I never got to sleep until 6. I am very familiar with latency. Thanks Gary!

  6. Question as I re-visit your blogs and try to “re-learn” the lessons from a couple years ago. I want to make sure I am interpreting this correctly and remembering our conversation correctly. While I understand the element that the consequence (be it a treat or a bonk) may be delayed and still be effective, the click or “no” is delivered quickly as possible when the behavior occurs? For example, Dallas loses his mind when the mailman walks by, but the Bonker is not within reach…… The “NO!” is delivered in conjunction with the behavior, but the bonk can come after the “no!” ? Same with a reward, click upon delivering the desired behavior even if a treat is not readily available? I understand what you’re saying about conditioning, especially with the mealtime example, but I’m not following how a consequence – positive or negative – can be associated with a behavior if there is a notable time delay. If Dallas barks at the mailman, I say “no”, but don’t bonk him until he’s walked away doesn’t the bonk get associated with a behavior other than the mailman?

    • Jackie,
      The click, or “NO”, defines the moment by ending information transfer. The presentation of the tangible event is almost meaningless if it is predictable. In a kennel situation, none of the dogs are fed immediately after hearing the crunch of a scoop dig into the feed bucket. Yet they all know the sound of the crunch, even though some of them have never been fed within five or ten minutes of the sound.

      I was at a client’s house the other day and timed it. From the moment I pressed the doorbell it was 38 seconds before the door was opened. They have a kiddie gate that keeps the puppy from getting close than about 15 feet from the guest. When the door opened, the puppy was jumping all over the kiddie gate. The pup had learned the connection between the doorbell and guests. The time was never less than ten seconds and most often closer to 45. This delay is entirely common. Longer times too – when I was in the National Guard, I often observed animals going to high ground when thunder clouds formed over the Superstitions. It would be hours before any flash flood could get there – yet the headed for high ground in a timely fashion.

  7. So, if I’m understanding what you are saying, if the click or “no” is delivered at the time of the behavior, the reward or bonk can be delivered at leisure?

Leave a Reply to Amber Ludwig Cancel reply

Your email address will not be published. Required fields are marked *