After the molar teeth of rats were extracted, the rats were fed with powdered food for 135 weeks. Although the performance in the radial arm maze was progressively acquired by daily training, an increase in the number of errors and a decrease in the initial correct responses were observed in the teethless aged rats compared to the control aged rats, indicating impaired acquisition of spatial memory in the teethless aged rats...the extracellular ACh level of the teethless aged rats under high-concentration of K+ and atropine sulfate stimulation was significantly low compared to that of the control aged rats. These results suggest that the impairment of spatial memory in the teethless aged rats may be due to the functional deterioration of the cholinergic neuronal system induced by tooth loss
It’s not a long paper. Skimming, the major problems I see:
the usual problems with animal studies: tiny sample size (9 in the control and 10 in the experimental, apparently), unclear randomization, no mentioned blinding of experimenters or raters
they didn’t show removing teeth caused lower performance, they showed removing teeth and feeding on a liquid diet caused lower performance. (On the plus side, they say they anesthesized both groups, so that removes a serious confound.)
The experimental group had its teeth removed & also was fed liquid, while the control group kept its teeth & also ate normal pellets. Hence, the decreased performance could’ve been caused (ignoring the issues of bias and sampling error) by either the removal of teeth, the liquid food, or some interaction thereof (perhaps liquid food aggravating tooth infection caused by the surgery?). They do say
Kawamura [6: “The effect of food consistency on conditioned
avoidance response in mice and rats”] has reported the relationship between mastication and learning and memory in young rats. He has also reported that rats fed with a powdered diet had poor results of learning and memory compared to those fed with a solid diet.
but I haven’t looked at it and in any case, given how much varies from lab to lab, this is a basic issue which needs to be verified in your own sample, and not just hope that it’s a universal. Also, if Kawamura finds that liquid food on its own damages learning & memory compared to a solid diet, how are you showing anything new by looking at liquid+surgery & finding damage...?
Their data is purely a post-comparison. They say they did the surgery, and then apparently left the rats alone for 135 weeks before doing the radial arm maze test.
So there’s no way to know what the decline looked like or when it happened. It’s perfectly possible that the toothless rats suffered a single sudden shock to their system from the surgery and that permanently degraded their memory, or that they had ongoing chronic inflammation or infection.
Worse, the difference may have been there from the start, they never checked. Randomization with such small n can easily fail to balance groups, that’s one reason for pre-tests: to verify that a difference in the groups on the post-test wasn’t there from the start but can be attributed to the experimental condition.
I’m not sure this can be described as a true ‘randomized experiment’. They never actually say that the selection of rats was random or how the animals were picked for their group, and there’s a weird pattern in the writing where they only ever write about the toothless rats being subjected to procedures even though logically you’d say stuff like ‘all the rats were tested on X’; eg:
After the molar teeth of rats were extracted, the rats were fed with powdered food for 135 weeks...Animals (11 weeks old) were anesthetized with sodium pentobarbital (40 mg/kg i.p.) and all maxillary and mandibular molars were extracted. Animals given anesthesia alone, without undergoing extraction of the molar teeth, were used as control aged rats...One hundred and thirty-five weeks after the surgery, the ability of learning and memory in the aged rats without molar teeth (hereafter referred to as ‘teethless’) was examined by using the radial arm maze [9], and compared to the control aged rats...Nine weeks after the learning and memory study, the ability of releasing ACh in the parietal cortex of teethless aged rats was examined by using in vivo microdialysis methods [5]...In order to examine the functional changes in cholinergic neuronal system of the teethless aged rats, animals were stimulated by high-concentration of K ÷ at 100mM or atropine sulfate at 3 ktM for 15 rain when the level of extracellular ACh stabilized.
Plus, Figure 1 reports 9⁄10 rats, but by Figure 2, we’re down to 5⁄5 rats. Huh? This makes me wonder if they’re reusing control rats from a previous experiment, or reusing their data, and only actually had experimental rats. (The use of “historical controls” is apparently not uncommon in animal research.)
This would massively compromise their results because rats change over time, litters of rats will correlate in traits like memory, and these effects are all large enough to produce many bogus results if you were to, say, take 10 rats from 1 litter as your control group and 9 rats from another litter as your experimental group. Just like with humans, one family of rats may have a very different average from another family. (See the very cool paper “Design, power, and interpretation of studies in the standard murine model of ALS”, Scott et al 2008, which helpfully notes on pg5 that when you have a mouse study with 10⁄10 mice similar to this study and the null is true, “an apparent effect [of >5% difference in survival] would be seen in 58% of studies”. Which really makes you think about a small difference in # of errors in maze performance.)
Their reward may have been a bit screwy in the memory task:
The apparatus was placed 40 cm above the floor. At the end of each arm there was a food cup that held a single 50-mg food pellet. Prior to the maze task, animals were kept on a restricted diet and the body weight was reduced to 80-85% of their normal weight over a 1-week period; water was freely available. Before the actual training began, the animals were allowed to explore the apparatus, for 10 min a day, for 2 days. For the following 16 trials, each animal was placed individually in the center of the maze and allowed to consume the bait in the food cup.
If this description is literally accurate, there’s a problem. They don’t mention the setup differing between groups! So this “food pellet” is the reward which gives the rats motivation to solve the maze… but you’ve removed the teeth from half the rats and can only feed them liquid. And you’re surprised the toothless rats perform worse? I’m reminded of the reward confounds in much animal intelligence research.
the authors mention excluding the other maze performance variable:
The teethless aged rats showed impairment performance during the acquisition of the radial arm maze task, as revealed by the increased number of errors (Fig. 1) and the decreased number of initial correct responses (data not shown).
One wonders if the # of initially correct responses would have reached p<0.05. Good old researcher degrees of freedom...
So overall, I would have to say this result seems to be extremely weak.
The most relevant part is probably another study mqrius mentions, “The effect of the loss of molar teeth on spatial memory and acetylcholine release from the parietal cortex in aged rats”, Kato et al 1997 (available through Libgen):
It’s not a long paper. Skimming, the major problems I see:
the usual problems with animal studies: tiny sample size (9 in the control and 10 in the experimental, apparently), unclear randomization, no mentioned blinding of experimenters or raters
they didn’t show removing teeth caused lower performance, they showed removing teeth and feeding on a liquid diet caused lower performance. (On the plus side, they say they anesthesized both groups, so that removes a serious confound.)
The experimental group had its teeth removed & also was fed liquid, while the control group kept its teeth & also ate normal pellets. Hence, the decreased performance could’ve been caused (ignoring the issues of bias and sampling error) by either the removal of teeth, the liquid food, or some interaction thereof (perhaps liquid food aggravating tooth infection caused by the surgery?). They do say
but I haven’t looked at it and in any case, given how much varies from lab to lab, this is a basic issue which needs to be verified in your own sample, and not just hope that it’s a universal. Also, if Kawamura finds that liquid food on its own damages learning & memory compared to a solid diet, how are you showing anything new by looking at liquid+surgery & finding damage...?
Their data is purely a post-comparison. They say they did the surgery, and then apparently left the rats alone for 135 weeks before doing the radial arm maze test.
So there’s no way to know what the decline looked like or when it happened. It’s perfectly possible that the toothless rats suffered a single sudden shock to their system from the surgery and that permanently degraded their memory, or that they had ongoing chronic inflammation or infection.
Worse, the difference may have been there from the start, they never checked. Randomization with such small n can easily fail to balance groups, that’s one reason for pre-tests: to verify that a difference in the groups on the post-test wasn’t there from the start but can be attributed to the experimental condition.
I’m not sure this can be described as a true ‘randomized experiment’. They never actually say that the selection of rats was random or how the animals were picked for their group, and there’s a weird pattern in the writing where they only ever write about the toothless rats being subjected to procedures even though logically you’d say stuff like ‘all the rats were tested on X’; eg:
Plus, Figure 1 reports 9⁄10 rats, but by Figure 2, we’re down to 5⁄5 rats. Huh? This makes me wonder if they’re reusing control rats from a previous experiment, or reusing their data, and only actually had experimental rats. (The use of “historical controls” is apparently not uncommon in animal research.)
This would massively compromise their results because rats change over time, litters of rats will correlate in traits like memory, and these effects are all large enough to produce many bogus results if you were to, say, take 10 rats from 1 litter as your control group and 9 rats from another litter as your experimental group. Just like with humans, one family of rats may have a very different average from another family. (See the very cool paper “Design, power, and interpretation of studies in the standard murine model of ALS”, Scott et al 2008, which helpfully notes on pg5 that when you have a mouse study with 10⁄10 mice similar to this study and the null is true, “an apparent effect [of >5% difference in survival] would be seen in 58% of studies”. Which really makes you think about a small difference in # of errors in maze performance.)
Their reward may have been a bit screwy in the memory task:
If this description is literally accurate, there’s a problem. They don’t mention the setup differing between groups! So this “food pellet” is the reward which gives the rats motivation to solve the maze… but you’ve removed the teeth from half the rats and can only feed them liquid. And you’re surprised the toothless rats perform worse? I’m reminded of the reward confounds in much animal intelligence research.
the authors mention excluding the other maze performance variable:
One wonders if the # of initially correct responses would have reached p<0.05. Good old researcher degrees of freedom...
So overall, I would have to say this result seems to be extremely weak.