Breakfast made six detection dogs choosier about odours that only resembled the target

Two error types moving independently is the finding here. A nose that has genuinely got worse shows it in both columns at once: fewer targets found and more non-targets called. These dogs held their hit rate at around 90% whether the session ran before or after the morning meal ^[5], which puts the miss rate near one target in ten either way ^[18], and leaves the fed-versus-fasted difference sitting in the false-alarm column ^[6].

Associate Professor Tim Edwards of Waikato reads that as a change in how likely a dog is to say the target is present, rather than a change in what it can smell ^[7][19]. His analogy is the urgent hunt for lost keys, when every small metal object starts to look like them ^[9]. The mechanism the team proposes is that hunger makes food a stronger motivator, and a dog that badly wants the reward responds more broadly to smells that resemble the target ^[10]. The dogs in the fasted sessions were working for treats ^[8], so the reward currency and the deprivation are the same substance.

The companion paper in Behavioural Processes does useful work on a duller explanation. Unfed dogs whined more and paid more attention to the food dispenser, but they stayed engaged with the scent task ^[16]. The dogs were still working, and still finding targets at the same rate, so the extra false alarms reflect something other than a checked-out animal.

What remains open is how large the effect gets in a working programme. Six dogs, one four-odour target mixture, an automated apparatus that keeps a handler's expectations out of the loop ^[3], and a public account that reports the drop in rejections as significant without giving the rates ^[21]. The authors flag considerable variation between individual dogs ^[15], which is exactly the term you would expect to dominate when the unit of deployment is one dog with one handler. They also say plainly that this is not an argument for a large meal immediately before every job, and that any feeding strategy has to account for the type and intensity of work, the individual dog, and how it is trained and rewarded ^[13]. Edwards puts translation to operational settings in the further-research column ^[14].

The practical read is narrow and real. If a programme's complaint is false alerts rather than misses, this study points at a variable operators already control at no cost, which is when the dog last ate. Co-author Dr Jade Fountain notes that in discrimination work a false alarm can matter as much as a missed target, particularly where people are relying on the dog to separate one scent from another ^[11], and the roles keep multiplying: conservation, disease detection, explosives and landmines ^[12]. Testing meal timing against hit and false-alarm logs you already keep is cheaper than buying retraining hours. The schedule itself still has to be earned locally, dog by dog ^[15].