Most of the answers thus far have focused on "intelligence" as being a spectrum or gauge, which is the most common definition. On that scale, we may consider sapience, as H. Beam Piper once wrote, as "a mental boiling point".
Most answers then consider how a species that has achieved sapience would end up reducing their average intelligence, but remaining sapient, staying above that boiling point identifiable by traits like a means of symbolic communication (speech/writing). While those answers are good at answering that reading of your question, there's another way to interpret that question, by considering "intelligence" as "the quantity of mental capability required for sentient thought" and thus a rough synonym of "sapience". Read that way, the question is, how can a species that has developed a sentient mental state lose that ability to think at a sentient level?
We have not, in the real world, ever seen this happen, at least not to our knowledge, because to our knowledge we are the only sapient species on this planet, and there have never been more than a handful of distinguishable (but at some level interrelatable) sapient primate species. However, our knowledge of the taxonomy of life on Earth is woefully incomplete; less than 1% of all species to have ever lived on the Earth are extant and thus available for behavioral study, and the fossil record is a notoriously poor source of clues as to a species' mentation (and that's of the species that leave a good fossil record).
So it's absolutely possible that some sauroid species in what's now the Yucatan Peninsula achieved some recognizable form of sapient thought, less than a million years before the nearby Chicxulub Impact wiped them off the map completely, fossil record and all. If the species that ascended to sapience were located further away from the impact site, it might have survived longer and/or their remains might be better preserved, so you'd think we'd have found a couple good fossil skeletal examples. Again, we think we know of less than 1% of all species ever to have lived on this planet, so the unknown space in which such a creature could have existed is fairly large.
Now, that leads us to a pretty good possibility. Sapience tends to require fairly high macronutrient intake for a given body size; that brain needs a lot of blood glucose to function properly. Mass extinction events like the Chicxulub Impact cause a pretty thorough collapse of the food web, and the species with the highest caloric needs often come out on the losing end pretty quickly. Climate change, asteroid impact, extinction of a key species (honeybees are a good candidate) are all really good ways for h. sapiens to find ourselves at an evolutionary dead-end.
Now, that's kind of cheating, because losing our ability to think because we're dead probably wasn't what you had in mind. However, if you consider our genetic relatives in the primate family as kindred enough, much as we consider modern-day reptiles and birds to be the closest descendants of the dinosaurs (however actually distant), it may well be that some marmoset in the tropical latitudes, on a land mass big enough to stay above the oceans' surface, may end up being the primate family's scion, that some future sapient octopus-derived species will describe as "the closest living relative to these far larger and more intelligent bipedal primates, whose fossil record is endemic on every land mass".
There is an important and closely-related point to make here; The definition of homo sapiens (and, we think, any other life form we identify as sapient) is defined as a species in large part based on that trait. If the human genetic line were, by any chain of events, to "devolve" to a sub-sapient mental state, in whole or part, any remaining sapient observer may well no longer consider those examples to be homo sapiens and therefore no longer "human". So there's somewhat of a tautology here; we're human because we're sentient, and if our genetic descendants lose that trait, they'd no longer be human.
Anyway, bringing this to an actual answer, it's kind of hard to envision an Earth populated with a genetic descendant of modern humans are no longer sentient, where that didn't happen due to an initial cataclysm that, ironically, makes life easier than before for those who survive the event. The cataclysm would have to be minimally damaging to the non-human ecosystem (otherwise the food web collapses, which definitely makes life harder and thus forces the remaining humans to have to think more), so asteroid impact is right out, and nuclear war is iffy, but possible if the theories that nuclear winter wouldn't happen are true. Pure climate change is going to make life harder for most as well.
The only such cataclysm I can think of would be pandemic, something that makes COVID look like hay fever. Contagion's fictional MEV-1, but even more contagious and deadlier (IIRC, MEV-1 had a mortality rate of 20-30% with an R0 of 4; something closer to 90% mortality with the same contagion rate would about do it), would literally decimate the world population, but for less advanced civilizations, isolated from the developed world and its deadly disease, life would actually get better in the next few hundred years after the virus had burned through the world's population. The severe cut in CO2 production and the dramatic release of pressure on game, fish and forest would, within just a few decades, produce an abundance of living biomass helping to sequester atmospheric CO2, not only reversing climate change but making the caloric needs of your tribe that much easier. If you don't have to think as much about where your next meal comes from, or coordinate as much with fellow humans to get it, you could, conceivably and eventually, arrive at a new steady-state where humans don't have to use their impressive brainpower anymore, and over many generations those abilities could atrophy, seeing the species regress to ice-age or earlier intelligence levels.
The biggest counterexample refuting such a theory is the Sentinelese people. The inhabitants of North Sentinel Island in the Indian Ocean are fiercely territorial, attacking any outsider who so much as sets foot on the island, such that there is no record of a successful, friendly communication between the Sentinelese and anyone else. There really is no talking to some people. In near-absolute isolation from the rest of the world, the Sentinelese are basically frozen in a neolithic state of development, one of very few known societies to be completely unpolluted by the last, oh, maybe 8,000 years or so of human mental, societal and technological progress. Yet despite all this, the Sentinelese are definitely sentient based on everything we have been able to observe at a distance. They have a spoken language, if not a written one, they know of fire and how to produce it, and they make force-altering machines, such as knives and bows and arrows.
While there are some glimmers of these traits among more intelligent animal species, there's no true comparison to be had except to ourselves; these are human beings, however primitive, and they're sapient. Thousands of years of human history passing them by has not changed that, and it's very unlikely thousands more will see much difference. They know what they need to know to survive, at least within their own isolated environment, and whatever folklore they pass down that keeps them so violently wary of outsiders has served them well, as most of the rest of the Andamanese people that populate the Andaman and Nicobar Islands have fared considerably worse against encroachment by non-indigenous cultures (technically part of the Indonesian archipelago, the Andaman and Nicobar Islands are an Indian possession and also close to Myanmar and Thailand).
So at the end of the day, having no more complicated a possession than a bow and arrow doesn't mean you have no inherent capability for sentient thought; quite the contrary, this level of technology is well inside our definition of sapience, and this knocks a big breach in the idea that the human race could ever "lose" our sapience. Merely not progressing isn't regressing.