In this article

1. Praise words in a flat voice
2. Word first, object second
3. Come, stay, outside: words for doing things
4. A nearly right word is not the same word
5. The habits that blur a cue
6. A routine for one clear cue
7. When an older dog stops answering

Say the word 'walk' in some homes and a dog who was fast asleep a second ago is spinning by the front door. It looks like understanding. But you probably also glanced at the lead, reached for your keys and let your voice lift, so which part did your dog actually hear?

Brain scanners and electrodes have started to answer that, though less tidily than the headlines suggested. Here is what the research shows, where it stops, and how to make your cues easier to follow.

Praise words in a flat voice

The study that made the headlines came from Budapest in 2016. The team's dogs are family pets, trained with rewards to lie awake and unrestrained in an MRI scanner for about eight minutes at a time².

They heard a trainer they knew say three praise words, the Hungarian for 'that's it', 'clever' and 'well done', and three words that mean nothing to a dog: 'as if', 'such' and 'yet'. Each word came in a warm, praising tone and in a flat, neutral one². The dogs' brains told the meaningful words from the meaningless ones whatever the tone was doing¹.

Tone still counted. The brain's reward areas became more active only when a praise word arrived in a praising voice¹. So the word and the warmth seem to work together.

A 2020 follow-up with 12 dogs, all but one from the first study, found the praise-word effect again². Even so, the researchers write that from their results alone they cannot claim the praise words meant something to the dogs: the sounds may just have been better learnt². What the work does show is that dogs learn about words, in a way that neither the sound of the recordings nor how often a dog hears each word can explain².

Word first, object second

The clearest sign that a word can point at a thing came from the same group in 2024, using electrodes instead of a scanner. Dogs heard the name of an object and were then shown an object: sometimes the one just named, sometimes a different one³.

When the object did not match, the dogs' brains showed a mismatch response between about 200 and 600 milliseconds after it appeared, close in timing to the N400, the signal human brains give when a word is paired with the wrong object³. The gap was bigger for words the owners said their dogs knew well, and it appeared whatever the size of a dog's vocabulary³.

This suggests more than a sound predicting something nice: a word may bring a particular thing to mind. The authors call it the first neural evidence for object word knowledge in any animal other than us³.

Come, stay, outside: words for doing things

Every word in that test named an object, but the words that send dogs to the door are mostly instructions and events. In 2025 the Budapest team analysed brain scans from 20 pet dogs listening to four everyday instructions in Hungarian ('come', 'go', 'stay' and 'lie down'), recorded in a friendly voice by four people the dogs did not know⁴.

Compared with nonsense sounds, the real words drew more activity in the brain's hearing areas, and also in areas involved in movement and its control⁴. The dogs were lying still with nothing to do, so the authors suggest the words were calling up what they mean, not just sounding familiar⁴.

Event words have been tested in a different way, with dogs whose owners had taught them to use soundboards, buttons that play a recorded word. At home, a researcher pressed the buttons for 'outside', 'play' and 'food' while the owner sat nearby in a sleep mask and noise-cancelling headphones; in a second version, owners pressed the button or said the word, then did not speak, point or react for a minute⁵.

Across both, dogs showed about seven times more door-directed behaviour after 'outside', and seven times more toy-directed behaviour after 'play'⁵. Food words drew no clear response, perhaps, the researchers suggest, because some dogs had already eaten or did not expect a meal at that time of day⁵. It made no difference whether a button or the owner's voice produced the word⁵. These were soundboard-trained dogs, though, and the authors want comparisons with dogs who have never used one⁵.

As for 'walk', we could not find a study of that word. It may work like 'outside', but for now that is our guess.

A nearly right word is not the same word

The 2025 study had a twist: alongside each real word, the dogs heard a near miss with one vowel swapped, so gyere ('come') became gyare. In most of the brain areas that responded to words, the near misses drew a response no different from the real thing, which the authors read as dogs storing the sound of a word more roughly than we do⁴.

Behaviour adds a caution. In a 2005 study, six dogs were trained to sit and come reliably to tape-recorded commands, then heard versions with some of the sounds changed. Their performance dropped markedly in every case⁶. The 2025 authors accept that when a dog has to act on a word, changes like these may affect the response⁴.

Together, the studies suggest a nearly right word may sound familiar to your dog without working as well as the real one. Keep each cue the same. Choosing clearly different words for different jobs is a sensible starting point too, though these studies did not test that advice.

Kind of word	What was found	What it suggests
Praise words	Brains told them from meaningless words, warm or flat1	Tone is not doing all the work
Object names	A brain mismatch signal when word and object differ3	Some words point at things
Instructions	Hearing and movement areas responded4	'Come' and 'stay' may call up their meaning
Outside and play	Soundboard dogs acted on them with no other cues5	Event words can carry the message
'Walk'	Not tested in any study we found	Still an open question

The habits that blur a cue

How we talk around a cue matters too. Researchers at the University of Lincoln filmed 56 handlers asking their dogs to sit during weekly classes at four training schools in England and Switzerland⁷. Dogs were less likely to sit when they were not looking at their handler, and when the handler said something else before the cue⁷. And only 61% of the handlers gave the spoken cue they had said they used⁷.

A controlled follow-up with 12 dogs tested this more closely. An unrelated word before the cue (the German word 'Banane') cut the response both to a cue the dogs knew well and to one they had just learnt. The dog's name straight before the cue made no difference, but the name followed by a two-second pause cut the response to the new cue⁷. In an unfamiliar room, the new cue lost ground whatever came before it, which the authors suggest is because a less established cue is harder to carry somewhere new⁷.

A routine for one clear cue

One job, one word

1. Choose the jobComing when called, say, with one word that sounds unlike your other cues
2. Write the word downSo everyone at home says the same one, every time
3. Get attentionYour dog's name is fine, as long as the cue follows straight away⁷
4. Say it onceIn your normal voice, then reward warmly the moment they get it right
5. Take it on tourA new cue may not travel at first⁷, so practise in other rooms, then the garden, then quiet places before busy ones
6. Test the word aloneSay it flatly with your hands still, or from another room

That last step lets you try the word on its own. The soundboard dogs answered 'outside' and 'play' with no other clues to go on⁵, so try a familiar cue flatly, with no glance at the lead. If your dog struggles, dogs follow spoken commands less well without a visual signal, as the Lincoln researchers note from earlier work⁷.

When an older dog stops answering

Hearing loss is common as dogs age, starting at around eight to ten years⁸. In a US study of 39 older dogs whose hearing was measured with a brain-response test, worse hearing went with higher scores on a dementia scale and lower owner ratings for vitality and companionship⁸. Owners had noticed the problem in 75% of the dogs with clear hearing loss⁸.

That study did not test cues, so this step is ours: if an older dog stops answering words they knew well, ask your vet to check their hearing. Go sooner if the change is sudden, because pain and illness can change how a dog responds too. The researchers suggest dogs may lean on visual cues to make up for some of the loss⁸, so a hand signal taught alongside each word gives an older dog a second way to follow you.

At any age, if a cue keeps falling apart, a qualified reward-based trainer can watch you and your dog together and spot what is getting in the way.

Your dog is sorting the sounds you make into the ones that matter and the ones that do not, and the evidence that some of them carry meaning keeps growing. Say the word, mean it, and use the same one tomorrow.

Sources and further reading8 sources

The small numbers in the article link to the source behind each fact. Sources open on their own websites.

1. Neural mechanisms for lexical processing in dogsScience (AAAS)
2. Multilevel fMRI adaptation for spoken word processing in the awake dog brainScientific Reports (Nature Portfolio)
3. Neural evidence for referential understanding of object words in dogsCurrent Biology (Cell Press)
4. Action instruction word processing in the dog brain entails both auditory form identification and meaning representationNeuroImage (Elsevier)
5. How do soundboard-trained dogs respond to human button presses? An investigation into word comprehensionPLOS ONE
6. The effect of human command phonetic characteristics on auditory cognition in dogs (Canis familiaris)Journal of Comparative Psychology (APA)
7. Factors affecting response of dogs to obedience instruction: a field and experimental studyApplied Animal Behaviour Science (Elsevier)
8. Relationship between hearing, cognitive function, and quality of life in aging companion dogsJournal of Veterinary Internal Medicine (ACVIM/Wiley)