Editing Metacognition (section)

==Animal metacognition==

=== In nonhuman primates ===

==== Chimpanzees ====
Beran, Smith, and Perdue (2013) found that chimpanzees showed metacognitive monitoring in the information-seeking task.<ref>{{Cite journal|last1=Beran|first1=Michael J.|last2=Smith|first2=J. David|last3=Perdue|first3=Bonnie M.|date=2013-03-18|title=Language-Trained Chimpanzees (Pan troglodytes) Name What They Have Seen but Look First at What They Have Not Seen|url= |journal=Psychological Science|volume=24|issue=5|pages=660–666|language=en|doi=10.1177/0956797612458936|pmc=3902479|pmid=23508741}}</ref> In their studies, three language-trained chimpanzees were asked to use the keyboard to name the food item in order to get the food. The food in the container was either visible to them or they had to move toward the container to see its contents. Studies shown that chimpanzees more often checked what was in the container first if the food in the container was hidden. But when the food was visible to them, the chimpanzees were more likely to directly approach the keyboard and reported the identity of the food without looking again in the container. Their results suggested that chimpanzees know what they have seen and show effective information-seeking behavior when information is incomplete.

==== Rhesus macaques (''Macaca mulatta'') ====
Morgan et al. (2014) investigated whether rhesus macaques can make both retrospective and prospective metacognitive judgments on the same memory task.<ref>{{Cite journal|last1=Morgan|first1=Gin|last2=Kornell|first2=Nate|last3=Kornblum|first3=Tamar|last4=Terrace|first4=Herbert S.|date=March 2014|title=Retrospective and prospective metacognitive judgments in rhesus macaques (Macaca mulatta)|url= |journal=Animal Cognition|language=en|volume=17|issue=2|pages=249–257|doi=10.1007/s10071-013-0657-4|issn=1435-9448|pmc=3883882|pmid=23812677}}</ref> Risk choices were introduced to assess the monkey's confidence about their memories. Two male rhesus monkeys (''Macaca mulatta'') were trained in a computerized token economy task first in which they can accumulate tokens to exchange food rewards. Monkeys were presented with multiple images of common objects simultaneously and then a moving border appearing on the screen indicating the target. Immediately following the presentation, the target images and some distractors were shown in the test. During the training phase, monkeys received immediate feedback after they made responses. They can earn two tokens if they make correct choices but lost two tokens if they were wrong.

In Experiment 1, the confidence rating was introduced after they completed their responses in order to test the retrospective metamemory judgments. After each response, a high-risk and a low-risk choice were provided to the monkeys. They could earn one token regardless of their accuracy if they choose the low-risk option. When they chose high-risk, they were rewarded with three tokens if their memory response was correct on that trial but lost three tokens if they made incorrect responses. Morgan and colleagues (2014) found a significant positive correlation between memory accuracy and risk choice in two rhesus monkeys. That is, they were more likely to select the high-risk option if they answered correctly in the working memory task but select the low-risk option if they were failed in the memory task.

Then Morgan et al. (2014) examine monkeys’ prospective metacognitive monitoring skills in Experiment 2. This study employed the same design except that two monkeys were asked to make low-risk or high-risk confidence judgment before they make actual responses to measure their judgments about future events. Similarly, the monkeys were more often to choose high-risk confidence judgment before answering correctly in working memory task and tended to choose the low-risk option before providing an incorrect response. These two studies indicated that rhesus monkeys can accurately monitor their performance and provided evidence of metacognitive abilities in monkeys.

=== In rats ===
In addition to nonhuman primates, other animals are also shown metacognition. Foote and Crystal (2007) provided the first evidence that rats have the knowledge of what they know in a perceptual discrimination task.<ref>{{Cite journal|last1=Foote|first1=Allison L.|last2=Crystal|first2=Jonathon D.|date=March 2007|title=Metacognition in the Rat|journal=Current Biology|volume=17|issue=6|pages=551–555|doi=10.1016/j.cub.2007.01.061|pmid=17346969|issn=0960-9822|pmc=1861845|bibcode=2007CBio...17..551F }}</ref> Rats were required to classify brief noises as short or long. Some noises with intermediate durations were difficult to discriminate as short or long. Rats were provided with an option to decline to take the test on some trials but were forced to make responses on other trials. If they chose to take the test and respond correctly, they would receive a high reward but no reward if their classification of noises was incorrect. But if the rats decline to take the test, they would be guaranteed a smaller reward. The results showed that rats were more likely to decline to take the test when the difficulty of noise discrimination increased, suggesting rats knew they do not have the correct answers and declined to take the test to receive the reward. Another finding is that the performance was better when they had chosen to take the test compared with if the rats were forced to make responses, proving that some uncertain trials were declined to improve the accuracy.

These responses pattern might be attributed to actively monitor their own mental states. Alternatively, external cues such as environmental cue associations could be used to explain their behaviors in the discrimination task. Rats might have learned the association between intermediate stimuli and the decline option over time. Longer response latencies or some features inherent to stimuli can serve as discriminative cues to decline tests. Therefore, Templer, Lee, and Preston (2017) utilized an olfactory-based delayed match to sample (DMTS) memory task to assess whether rats were capable of metacognitive responding adaptively.<ref>{{Cite journal|last1=Templer|first1=Victoria L.|last2=Lee|first2=Keith A.|last3=Preston|first3=Aidan J.|date=2017-07-01|title=Rats know when they remember: transfer of metacognitive responding across odor-based delayed match-to-sample tests|url= |journal=Animal Cognition|volume=20|issue=5|pages=891–906|doi=10.1007/s10071-017-1109-3|pmid=28669115|issn=1435-9448|pmc=5709207}}</ref> Rats were exposed to sample odor first and chose to either decline or take the four-choice memory test after a delay. The correct choices of odor were associated with high reward and incorrect choices have no reward. The decline options were accompanied by a small reward.

In experiment 2, some “no-sample” trials were added in the memory test in which no odor was provided before the test. They hypothesized that rats would decline more often when there was no sample odor presented compared with odor presented if rats could internally assess the memory strength. Alternatively, if the decline option was motivated by external environmental cues, the rats would be less likely to decline the test because no available external cues were presented. The results showed that rats were more likely to decline the test in no-sample trials relative to normal sample trials, supporting the notion that rats can track their internal memory strength.

To rule out other potential possibilities, they also manipulated memory strength by providing the sampled odor twice and varying the retention interval between the learning and the test. Templer and colleagues (2017) found rats were less likely to decline the test if they had been exposed to the sample twice, suggesting that their memory strength for these samples was increased. Longer delayed sample test was more often declined than short delayed test because their memory was better after the short delay. Overall, their series of studies demonstrated that rats could distinguish between remembering and forgetting and rule out the possibilities that decline use was modulated by the external cues such as environmental cue associations.

=== In pigeons ===
Research on metacognition of pigeons has shown limited success. Inman and Shettleworth (1999) employed the delayed match to sample (DMTS) procedure to test pigeons’ metacognition.<ref>{{Cite journal|last1=Inman|first1=Alastair|last2=Shettleworth|first2=Sara J.|date=1999|title=Detecting metamemory in nonverbal subjects: A test with pigeons.|url=http://dx.doi.org/10.1037/0097-7403.25.3.389|journal=Journal of Experimental Psychology: Animal Behavior Processes|volume=25|issue=3|pages=389–395|doi=10.1037/0097-7403.25.3.389|issn=1939-2184|url-access=subscription}}</ref> Pigeons were presented with one of three sample shapes (a triangle, a square, or a star) and then they were required to peck the matched sample when three stimuli simultaneously appeared on the screen at the end of the retention interval. A safe key was also presented in some trials next to three sample stimuli which allow them to decline that trial. Pigeons received a high reward for pecking correct stimuli, a middle-level reward for pecking the safe key, and nothing if they pecked the wrong stimuli. Inman and Shettleworth's first experiment found that pigeons’ accuracies were lower and they were more likely to choose the safe key as the retention interval between presentation of stimuli and test increased. However, in Experiment 2, when pigeons were presented with the option to escape or take the test before the test phase, there was no relationship between choosing the safe key and longer retention interval. Adams and Santi (2011) also employed the DMTS procedure in a perceptual discrimination task during which pigeons were trained to discriminate between durations of illumination.<ref>{{Cite journal|last=Allison Adams|first=Angelo Santi|title=Pigeons exhibit higher accuracy for chosen memory tests than for forced memory tests in duration matching-to-sample|journal=[[Learning & Behavior]]|year=2011|volume=39|issue=1|pages=1–11|doi=10.1007/s13420-010-0001-7|pmid=21264568|s2cid=40008821 |doi-access=free}}</ref> Pigeons did not choose the escape option more often as the retention interval increased during initial testing. After extended training, they learned to escape the difficult trials. However, these patterns might be attributed to the possibility that pigeons learned the association between escape responses and longer retention delay.<ref>{{Cite journal|last1=Iwasaki|first1=Sumie|last2=Kuroshima|first2=Hika|last3=Fujita|first3=Kazuo|date=2019-11-01|title=Pigeons show metamemory by requesting reduced working memory loads.|journal=Animal Behavior and Cognition|volume=6|issue=4|pages=247–253|doi=10.26451/abc.06.04.04.2019|issn=2372-5052|doi-access=free}}</ref>

In addition to DMTS paradigm, Castro and Wasserman (2013) proved that pigeons can exhibit adaptive and efficient information-seeking behavior in the same-different discrimination task.<ref>{{Cite journal|last1=Castro|first1=Leyre|last2=Wasserman|first2=Edward A.|date=March 2013|title=Information-seeking behavior: exploring metacognitive control in pigeons|url=http://link.springer.com/10.1007/s10071-012-0569-8|journal=Animal Cognition|language=en|volume=16|issue=2|pages=241–254|doi=10.1007/s10071-012-0569-8|pmid=23065186|s2cid=9730580|issn=1435-9448|url-access=subscription}}</ref> Two arrays of items were presented simultaneously in which the two sets of items were either identical or different from one another. Pigeons were required to distinguish between the two arrays of items in which the level of difficulty was varied. Pigeons were provided with an “Information” button and a “Go” button on some trials that they could increase the number of items in the arrays to make the discrimination easier or they can prompt to make responses by pecking the Go button. Castro and Wasserman found that the more difficult the task, the more often pigeons chose the information button to solve the discrimination task. This behavioral pattern indicated that pigeons could evaluate the difficulty of the task internally and actively search for information when is necessary.

=== In dogs ===
Dogs have shown a certain level of metacognition that they are sensitive to information they have acquired or not. Belger & Bräuer (2018) examined whether dogs could seek additional information when facing uncertain situations.<ref>{{Cite journal|last1=Belger|first1=Julia|last2=Bräuer|first2=Juliane|date=2018-11-12|title=Metacognition in dogs: Do dogs know they could be wrong?|journal=Learning & Behavior|volume=46|issue=4|pages=398–413|doi=10.3758/s13420-018-0367-5|pmid=30421122|pmc=6276073|issn=1543-4494|doi-access=free}}</ref> The experimenter put the reward behind one of the two fences in which dogs can see or cannot see where the reward was hidden. After that, dogs were encouraged to find the reward by walking around one fence. The dogs checked more frequently before selecting the fence when they did not see the baiting process compared with when they saw where the reward was hidden. However, contrary to apes,<ref>{{Cite journal|last=Call|first=Josep|title=Do apes know that they could be wrong?|url=https://pubmed.ncbi.nlm.nih.gov/20306102/|journal=[[Animal Cognition]]|year=2010|volume=13|issue=5|pages=689–700|doi=10.1007/s10071-010-0317-x|pmid=20306102|s2cid=14856244}}</ref> dogs did not show more checking behaviors when the delay between baiting the reward and selecting the fence was longer. Their findings suggested that dogs have some aspect of information-searching behaviors but less flexibly compared to apes.

=== In dolphins ===
Smith et al. (1995) evaluated whether dolphins have the ability of metacognitive monitoring in an auditory threshold paradigm.<ref>{{Cite journal|last1=Smith|first1=J. David|last2=Schull|first2=Jonathan|last3=Strote|first3=Jared|last4=McGee|first4=Kelli|last5=Egnor|first5=Roian|last6=Erb|first6=Linda|date=1995|title=The uncertain response in the bottlenosed dolphin (Tursiops truncatus).|url=http://dx.doi.org/10.1037/0096-3445.124.4.391|journal=Journal of Experimental Psychology: General|volume=124|issue=4|pages=391–408|doi=10.1037/0096-3445.124.4.391|pmid=8530911|issn=1939-2222|url-access=subscription}}</ref> A bottlenosed dolphin was trained to discriminate between high-frequency tones and low-frequency tones. An escape option was available on some trials associated with a small reward. Their studies showed that dolphins could appropriately use the uncertain response when the trials were difficult to discriminate.

=== Debate ===
There is consensus that nonhuman primates, especially great apes and rhesus monkeys, exhibit metacognitive control and monitoring behaviors.<ref>{{Cite journal|last1=Smith|first1=J. David|last2=Couchman|first2=Justin J.|last3=Beran|first3=Michael J.|date=2014|title=Animal metacognition: A tale of two comparative psychologies.|url= |journal=Journal of Comparative Psychology|volume=128|issue=2|pages=115–131|doi=10.1037/a0033105|pmid=23957740|issn=1939-2087|pmc=3929533}}</ref> But less convergent evidence was found in other animals such as rats and pigeons.<ref>{{Cite journal|last=Beran|first=Michael|date=2019-11-01|title=Animal metacognition: A decade of progress, problems, and the development of new prospects.|journal=Animal Behavior and Cognition|volume=6|issue=4|pages=223–229|doi=10.26451/abc.06.04.01.2019|issn=2372-5052|doi-access=free}}</ref> Some researchers criticized these methods and posited that these performances might be accounted for by low-level conditioning mechanisms.<ref>{{Cite journal|last1=Smith|first1=J. David|last2=Zakrzewski|first2=Alexandria C.|last3=Church|first3=Barbara A.|date=2015-12-15|title=Formal models in animal-metacognition research: the problem of interpreting animals' behavior|journal=Psychonomic Bulletin & Review|volume=23|issue=5|pages=1341–1353|doi=10.3758/s13423-015-0985-2|pmid=26669600|pmc=4909597|issn=1069-9384|doi-access=free}}</ref> Animals learned the association between reward and external stimuli through simple reinforcement models. However, many studies have demonstrated that the reinforcement model alone cannot explain animals’ behavioral patterns. Animals have shown adaptive metacognitive behavior even with the absence of concrete reward.<ref>{{Cite journal|last1=Beran|first1=Michael J.|last2=Smith|first2=J. David|last3=Coutinho|first3=Mariana V. C.|last4=Couchman|first4=Justin J.|last5=Boomer|first5=Joseph|date=2009|title=The psychological organization of "uncertainty" responses and "middle" responses: A dissociation in capuchin monkeys (Cebus apella).|url= |journal=Journal of Experimental Psychology: Animal Behavior Processes|volume=35|issue=3|pages=371–381|doi=10.1037/a0014626|pmid=19594282|issn=1939-2184|pmc=3901429}}</ref><ref>{{Cite journal|last1=Smith|first1=J. David|last2=Redford|first2=Joshua S.|last3=Beran|first3=Michael J.|last4=Washburn|first4=David A.|date=2009-06-13|title=Rhesus monkeys (Macaca mulatta) adaptively monitor uncertainty while multi-tasking|url= |journal=Animal Cognition|volume=13|issue=1|pages=93–101|doi=10.1007/s10071-009-0249-5|pmid=19526256|issn=1435-9448|pmc=3951156}}</ref>