How do you fix a dog that anticipates the next command before you give it?

Practicing sit followed immediately by down, in that exact order, across many repetitions teaches a dog a chain of behaviors rather than two independently cued ones, and this pattern can form within as few as 15 or 20 repeated sessions run the same way each time.

Asking for down, then sit, then down again, then adding a stay or a touch cue in between, in a genuinely unpredictable order breaks the chain by forcing the dog to actually listen for each specific word rather than run the familiar routine.

If a dog drops into a down before being asked, calmly asking it to sit back up and starting the sequence again, rather than accidentally rewarding the guess, makes it clear that only the behavior actually cued earns the reward.

Inserting a completely unrelated action, like touch or a simple wait, immediately after the sit and before any chance of an automatic down, forces the dog to actually process each new instruction individually rather than sliding straight into the familiar second half of an old, well worn sequence. Doing this for several sessions in a row is usually enough to break even a strongly established chain, especially one that has been practiced the same way for months, and the dog's pause before responding is a good sign the habit is finally loosening.