Dog owners: to pet or not to pet during thunderstorms or fireworks.

Revised August 2019.

Depending on who you ask, you get a vastly different answer to the question of whether it’s a good idea to pet a frightened dog (or other animal with whom you have an established relationship and that enjoys your touch) during noisy (scary) events.

In the scientific literature, the overwhelming majority of articles (if not all…?) recommend not petting, meaning that petting fearful dogs inadvertently reinforces the anxious behaviour.

Still, in the dog training community, many professional dog trainers do recommend petting an anxious animal, and it has been argued that “petting doesn’t reinforce fear, it reduces it”.

Does petting reduce fear – or reinforce it?

So which is it? Could both sides have a point?

Let’s start with terminology.


Does petting reinforce fear?

Well, the process of reinforcement occurs during operant learning. In positive reinforcement learning, animals learn that certain behaviours have desired outcomes, as they result in getting access to something the animal wants.

The animal performs a behaviour, receives reinforcement, and repeats the behaviour in order to get some more goodies.

Fear isn’t a behaviour, though, it’s an emotional state. 

And I sometimes hear people say “emotions can’t be reinforced – only behaviours can be reinforced”. That sweeping statement suggests that emotions have nothing to do with the process of reinforcement.

I disagree.

I think emotions have everything to do with reinforcement.

After all, without emotional responses to stimuli, those stimuli wouldn’t be reinforcing – or punishing.

Without emotional responses to incoming stimuli, those stimuli would remain neutral.

It’s because they elicit emotional reactions that those stimuli end up being reinforcers or punishers.

But – positive and negative emotions are regulated differently, by positive versus negative feedback loops.

Let’s see if I can unpack that last statement, because it’s  crucial to this discussion.

Emotions involved in positive reinforcement are in a positive feedback loop.

Imagine for a second a dog who sits, gets petted, and then sits some more.

Petting feels good (assuming it’s a reinforcer in the first place when delivered from one particular person to a particular animal, but let’s make that assumption, for now. People often overestimate how much their animals enjoy being petted, but that’s a topic for another blog post, perhaps…)

For a cuddle-seeking animal, petting is a positive reinforcer that leads to more behaviour (in this case, sitting) and a corresponding feel-good emotion: a positive feedback loop.

Emotions involved in positive reinforcement are in a positive feedback loop: behaviour serves to get access to MORE of the resource, and the animal strives to remain in or increase that pleasant emotional state.

The animal continues offering sits to get more attention and body contact and remain in that happy mood, at least until something more interesting happens – or until they satiate.

What about fear, then?

We know animals don’t like being fearful, but even behaviours that animals don’t appreciate may be reinforced if the reinforcer is valuable enough.

So, the fear-is-reinforced-by-petting argument goes: “the animal acts fearful, gets petted and then acts even more fearful in order to get petted some more.”

It is assumed that through positive reinforcement (adding something the animal enjoys), fearful behaviour will increase. And presumably, the accompanying emotion too. No distinction is usually made between observable behaviour and the corresponding emotional state.

However, fear is not comparable to operant behaviour such as sitting, coming when called or giving a high five.

Fear is an emotional state, and one that has immense survival value.

Fear leads to predictable observable behavioural reactions such as freezing, fleeing or (if cornered), fighting.

In other words, many behaviours associated with the emotion fear serve to reduce that emotional state, by removing or increasing the distance to the frightening stimulus.

This involves, by the way, the process of negative reinforcement (removing something the animal dislikes when the animal performs the correct behaviour).

For a frightened animal, hiding reduces the impact of the negative reinforcer, that scary thing.

In other words, it’s a negative feedback loop that reduces fear. So even though the observable behaviour (hiding) may be reinforced and showed again when the fearful stimulus is presented, the corresponding emotion is not.

Successful negative reinforcement leads to relief. A reduction of fear. 

Emotions involved in negative reinforcement are in a negative feedback loop.
Emotions involved in negative reinforcement are in a negative feedback loop: behaviour serves to DIMINISH the aversive stimulation, and the animal strives to reduce or terminate that unpleasant emotional state.

So, it’s hard to even imagine that you could positively reinforce the emotion of fear and the associated observable behaviours, such as hiding, since they work in a negative feedback loop.

Sitting produces treats when you’re hungry through positive reinforcement, and happy excitement increases.

In contrast, hiding eliminates aversive stimuli when you’re fearful through negative reinforcement, and fear is reduced.

Many fearful overt behaviours serve to decrease fear through negative reinforcement: an outcome the animal enjoys. Assuming petting is something else that the animal enjoys, it should have the same effect as hiding: to reduce fear.

So, the argument that petting-reinforces-fear is just plain incorrect since fear-related behaviours are governed by negative reinforcement that serves to remove the aversive stimulus, reducing the underlying emotional state through a negative feedback loop. Once fear has diminished beyond a certain level, the animal will stop hiding.

Hiding during a thunderstorm is negatively reinforced since it reduces fear. Likewise, soliciting petting from the owner is negatively reinforced since it reduces fear. Yes, your dog will seek you out to ask for comfort when frightened, just as he will attempt to hide, since both behaviours work to reduce fear!

So, many dog trainers will pet their frightened dog to good effect. However, they may be hesitant to recommend this procedure to someone who’s not so used to reading their dog and changing their own behaviour.

In my opinion, the people that warn that petting-could-make-fear-worse are not wrong, they just use the wrong terminology.

Petting may, unfortunately, lead to increased fear or more suffering; just not through the process of reinforcement.

I see no less than three potential mechanisms why this might occur.

Three ways that petting a frightened animal could make things worse: extinction bursts, sensitization and respondent conditioning.
  • Situation 1) The animal is really frightened, seeks attention and gets petted, which under normal conditions reduces arousal and feels really nice. But if the animal is too fearful, it could be that this effect is not achieved and high arousal is maintained. If petting no longer “works”, the animal is no longer being reinforced.

What happens when previously reinforced behaviour no longer works?

Extinction bursts, that’s what.

It is not uncommon that frightened dogs start climbing and pawing their owners, getting even more aroused when they get petted – perhaps because they’re not getting the relaxation they’ve learned to expect from being-petted-when-in-a-positive-or-neutral-mood.

So, hypothetically, at some arousal threshold, petting will no longer reduce arousal, and this may cause an extinction burst and increased arousal in frightened animals.

To reduce this risk, block out offensive stimuli such as sounds and sights by providing a good hiding place. Theoretically this might keep the arousal level below the threshold so that petting still “works”.

  • Situation 2) The animal is really frightened and has become sensitized to other stimuli, including touch.

Sensitization has multimodal effects, meaning that the other senses tend to become sensitized too, so touch may become aversive. And even though the animal may not become more fearful, petting may add another unwanted stimulus on top of everything else, and increase discomfort.

To reduce the risk of sensitization, again, block out the aversive stimuli.

  • Situation 3) Through respondent conditioning, fear may increase – or decrease.

Suppose that after each loud sound, you pet the animal. The animal thus learns that awful sounds are followed by petting, and through counterconditioning, the sounds become less aversive.

respondent - counterconditioning
Noise becomes a predictor of petting, something to look forward to.

However, what happens if there is a new loud sound immediately after petting occurred? Then it’s hard to predict which type of respondent conditioning will occur.

If you’re lucky, counterconditioning to noise occurs (noise predicting petting). Fear should decrease.

If you’re unlucky, counterconditioning occurs to petting (petting predicting noise). Fear might increase.

Petting-during-noise might increase fear, not through operant conditioning and reinforcement, but through respondent conditioning.

respondent - which type
When petting becomes a predictor of aversive things, the animal may learn to fear petting.

Incidentally, with the same reasoning, animals may learn to fear wonderful treats, too.

During a fireworks display, there’s typically more or less continuous sound, whereas during thunderstorms, annoying visual and auditory stimuli are less frequent.

I would thus expect that the risk of inadvertent respondent conditioning resulting in petting-increasing-fear would be higher during fireworks than during thunderstorms (unless the animal is fearful of the changes in barometric pressure or electrical discharges which may be a lot more continuous).

Do you agree – do dogs respond better to petting during thunder than fireworks? What’s your experience? Let us know in the comments’ section!

I would also expect that the intensity of the stimuli would affect the outcome of frequent-noises-interspersed-with-petting. If noise is loud, it may overshadow petting so that the animal doesn’t even notice that it’s being petted (unless it’s sensitized, which you’ll notice).

Sorry, if you’re unfamiliar with the term overshadowing, it means that one of the two stimuli is so obvious that the other stimulus isn’t noted.

If noise is soft, petting may overshadow the noise so that the animal doesn’t even notice that there is a storm outside… and if you’re using treats, make sure that they’re absolutely fabulous so that treats overshadow noise rather than the other way around.

Of course, another factor that will influence the outcome of the interaction is that the owner may be calmed by petting the dog – and that may have a huge influence on the animal!


Be cautious of petting a frightened dog during a noisy event (especially fireworks). It may make both of you feel better, but pay attention to the animal.

Stop if you see that petting makes things worse (which could occur through three different mechanisms). Above all, block the annoying stimuli as best you can: allow the animal to hide in a sound-proof place. If you’re giving treats after each noise, make sure they’re superbly yummy.

BTW, here’s everything else you need to know about how to reduce fear of noise in dogs during the acute phase, the off-season phase, as well as the preventive phase. Want it as an E-book? 😉

21 Replies to “Dog owners: to pet or not to pet during thunderstorms or fireworks.”

  1. I think that what you should do or not has a lot to do with the relationship you have with the dog in question. I have seven year old dog, Selma, who is really bothered by loud bangs such as fireworks and gunshots. We moved to a new place last year, with an armybase nearby – and got a lot of good training during the fall preparing for new years eve. Unfortunately she got traumatized during Christmas when some stupid kids started shooting fireworks in the subway when we stood waiting – all training lost! Since then it’s been a steep uphill climb to get things a bit sorted again – progress is slow but we’re getting there. Anyway … a couple of weeks back I started thinking: “if we hear a big bang when we’re outside I know that Selma will react to the sound with, at the least, moderate fear. Since Selma and I are very close I also know that she will notice that I’m perfectly aware of her feelings, so trying to just skip over her reactions would be useless acting. So … I will do the opposite”. So what I started doing was, when we hear a bang, I go up to her, sit calmly by her side holding her gently but firmly and tell her that I see and understand her fears but that we together will get through this – I actually say this to her, but it’s obviously not the words but the calm voice and me being present that does the trick (but the words mean something to me, putting me in a good mood). And it works! Selma is the kind of dog that remembers and anticipate things, and fears from a loud bang can hold on for hours – but with our new routine it wears of really fast and we have so much more space to work with her anxietys. So – if I should come to some kind of point here – petting and being near an anxious dog can work great … if you know the dog and the dog knows you enough to put some trust in your relationship.

    1. Thank you for sharing your story, Kaj! You’re making several important points: each dog is an individual, our own emotional state matters – and we can help the animal regulate autonomic arousal by remaining calm and collected..! 🙂

  2. Great article! I think it is precisely because not everyone is able to correctly judge whether or not she/he should pet a dog when it’s scared, the Reimer publication (2020) shows that interaction with a dog (talking, stroking) is not associated with any improvement or deterioration in the level of fear in dogs over time

    1. Ah, interesting; would be interesting to know whether momentarily successfully “soothing-a-dog-through-petting” has any long-term effects or not. (and not, as you suggest, confounded by ineffective petting)

    2. I know this is probably heresy (and there is little as uncommon or unreliable as ‘common sense’) but I think we do make this more complicated than it needs to be. We have probably all felt significant fear, and we have all probably comforted someone who is experiencing fear. So we know first hand that certain things are essential to whether comfort helps in the moment: for instance, a sense of security or safety in that person’s presence, whether through their demeanour or our relationship – do we trust them? Also whether there are things we can do to manage or cope with the fearful stimulus – passivity is harder to manage than action. So for some dogs burrowing into their nest will bring more comfort than being handled; but equally for some dogs going to their guardian may be their proactive response and so that very proactivity is a coping mechanism, and brings relief regardless of the petting itself.
      I have never known anyone, who after crying on my shoulder, came over to cry on me again simply because they wanted the cup of tea and the attention. I’ve never had an animal who trembled or cried or hid simply so that I would pet them, despite having many who have been scared of fireworks, and comfort always being on offer. When the fireworks stop, those behaviours stop. Petting is still available – so why would an animal use such tainted behaviours to ask for attention? Those behaviours – trembling, crying, hiding, are associated with fear, with something they hate. Attention is given in many other positive and non-conditional ways. I see no reason why those behaviours would become more likely outside of a fear response.
      And the fear itself – yes, it could become generalised to the room, the house, the owner. A dog may build a trigger picture with many elements in it. Yet we never say: Don’t feed your dog on November 5th or they will become fearful of food. Don’t say their name or their name will become tainted by the association with fireworks. Only don’t pet them.
      Given that our dogs presumably have many other points at which they receive affection, attention and petting, it seems unlikely that they would rehearse fear-related behaviours, which must have a profoundly negative association for them, in order to derive attention they can acquire much more pleasantly and simply in other ways.

  3. To reduce separation anxiety, I should be subtle about leaving a dog but make a big song and dance when I return. Would this mean the dog would want me to go due to the reward of me coming back?

    1. Graham, a relevant question! Firstly, I think that whether to make a big song-and-dance on returns is a bit debated, and it seems to me that successful separation anxiety training is more about really granular systematic desensitization, and less about counterconditioning. Secondly, it seems that what actually happens is that they go from petrified to resigned.

  4. I have a client (dog walking) who is fearful of noise. Let me set the stage. They live behind a train track. Not only does he feel the vibration of the train coming but the blowing of the horn as it come to the train crossing that is very close to their home. This track also has an area where they perform repairs and such behind their home. Flynn is a Cairn Terrier. He is a robust feisty dog. On his group walks away from his home he is generally himself. However when we are at certain trails he can hear chainsaws across the river or off in the distance. He is very fearful of noise. He will bolt and run to safety which is my van. He knows the trail and the way there. I am very aware of his behaviour and emotional states prior to this happening and will leash him up right away. Unfortunately I have 4 or 5 more dogs that I may or may not have to continue the walk, or return to the van. He is very treat motivated. So I am I gathering that if he comes to me for comfort and or I treat him in this state I am not reinforcing the emotional state he is in. The owner is very aware of his noise sensitivity.

  5. I would believe that it is not only the actually petting that will influence on how the dog reacts on it. The way the owner pets the dog and the relationship between the owner and the dog will also influence the dogs emotional reaction to the petting. If the dog trust the owner and the owner is relaxed and pets the dog calmly, I would belive that the petting could help the dog to calm down.

  6. I have a client whose 7-year old Swiss Shepherd dog used to be apprehensive of thunder (not completely fearful). In December during a thunderstorm, he was alone at home (without the other dogs) and since then he has started to hide in the bathroom whenever there is thunder. This has generalised to rain without thunder. He is now starting to hide in the bathroom when his owners go to work as well. The owners would prefer him not to be in the part of the house where the bathroom is because that is the cats’ domain. Would it help to create another safe space for him (like a crate) that can be placed in another part of the house, while working on SD/CC and the enrichment activities? I don’t want him to practice the behaviour for too long.

    1. Having many safe spaces would certainly be a good idea, but I wouldn’t confine the dog to the crate when the owners go away… also, perhaps some medication might be tried while they’re working through this?

  7. I thing that it is one thing for an owner to pet the dog but they also talk to them an most of time it makes it worst for the dog. As a trainer I feel that it is better for the dog to have safe place with a great treat to keep him busy, is a safer way to handle it. If they were be able to have the right body language themselves, keep their mood calm and stay relaxed, with very little talking to the dog then petting may work. If the owner has a clear understanding of what dog’s fear is or are but again that is very often not the case. That is why it is not easy to recommend that the pet their dog. At least this is my opinion.

    1. Hah, yes! I actually challenged dog owners to use different tones of voice in my last blog post just to illustrate this… 🙂

  8. Great article!
    This article is centered on the conditioning factors, it would be very fun to read some on the physiological factors that can occur when petting a dog during fearful moments, how it affects the dog and how it can be used in training.

    1. Hi, an interesting idea! I would guess that there is a “war of chemicals” going on that influences the threshold point discussed above. If petting “works”, oxytocin, endorphins etc are released and leads to that warm fuzzy feeling. If the animal is fearful ands stressed, I’d expect either a cortisol or an adrenaline/noradrenaline rush. At some point I assume that one system mutes the other…! Or rather, that in that warm fuzzy mood, the animal is less likely to pay attention to incoming stimuli that might signal danger. Once the danger-system is alerted it’s no wonder that petting no longer works: from the evolutionary perspective one can never afford to ignore danger signals…

Leave a Reply

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