In:
Proceedings of the National Academy of Sciences, Proceedings of the National Academy of Sciences, Vol. 108, No. 31 ( 2011-08-02)
Abstract:
Collectively, our findings thus demonstrate the ramifications of behavioral optimization, identify some of the critical neural determinants of this process, and serve as an important step in forging more transparent connections between the human and animal literatures on learning and memory. Spontaneous revisitation is therefore a robust example of ongoing behavioral optimization in humans. Revisitation was highly prevalent in neurologically intact individuals, and was associated with superior learning. Amnesic individuals exhibited relative failures to engage in revisitation and to benefit from it, showing the necessary role of the hippocampus in ongoing behavioral control. Furthermore, brain imaging in neurologically intact individuals suggested that the hippocampus did not implement revisitation alone, but participated with regions of frontal cortex and cerebellum involved in planning and prediction functions relevant to behavioral optimization. The specificity of these effects to self-directed engagement of spontaneous revisitation was demonstrated by their general absence in response to revisitation viewed during the passive condition. Although the effects we described are similar to some adaptive behaviors studied in nonhuman animals, the level of specificity provided by the passive condition is difficult to achieve in animals, in which it is difficult to dissociate the control of behavioral optimization from the behaviors themselves. We next sought to identify the larger brain circuitry related to spontaneous revisitation by examining activity in healthy, college-age individuals by using functional MRI ( n = 12; data unrelated to revisitation originally reported in ref. 1 ). Volitional study blocks and passive study blocks were scored according to the prevalence of spontaneous revisitation, and brain activity associated with revisitation was thus identified separately for these study conditions ( Fig. P1 D ). In the volitional condition, spontaneous revisitation was associated with activity in the hippocampus, consonant with our findings in amnesic individuals. In addition, regions of medial prefrontal cortex and cerebellum were identified, in opposite hemispheres. This crossed anatomical arrangement is notable, given that closed-loop frontocerebellar circuits cross the midline, and these circuits are thought to be involved in “executive control” functions ( 4 ), which include behavioral planning and prediction of the outcomes of behavior. Findings from the passive condition showed that this activity was specific to self-generation of revisitation, in that revisitation during passive viewing was not associated with any of these effects, and was instead related to simple habituation effects in rostral orbitofrontal cortex, likely caused by the repeated viewing of the same objects after brief delays. These findings indicate that the hippocampus is necessary for the ongoing behavioral optimization involved in spontaneous revisitation. Interestingly, amnesic individuals showed memory performance that was better than chance when tested after studying an entire set of objects, yet nonetheless were significantly impaired at engaging in revisitation while studying sets of objects. This suggests that hippocampal damage may impair the use of memory processing in the service of ongoing behavioral optimization—an impairment that can be observed on a timeframe shorter than long-term memory deficits associated with damage to this structure. Intriguingly, lesions made to the rodent hippocampus reduce the prevalence of an exploration strategy involving a back-and-forth viewing pattern highly similar to the revisitation pattern we describe here ( 3 ), suggesting that the role of the hippocampus in behavioral optimization may be common to mammals. However, it is also possible that hippocampal processing is necessary for short-term behavioral optimization, given the intuitive utility that memory would serve in identifying the most critical information during learning. Indeed, we found in two experiments that amnesic individuals engaged in spontaneous revisitation during volitional study significantly less than did comparison individuals, who were matched in age, sex, and other factors to the amnesic individuals, but without brain damage ( n = 3 amnesic individuals in the first experiment; data unrelated to revisitation originally reported in ref. 1 ; n = 4 amnesic individuals in the second experiment). Averaged across the two experiments, 37% of object-to-object transitions were part of spontaneous revisitation sequences in comparison individuals (nearly identical to results from college-age individuals described earlier), whereas this value was only 10% in amnesic individuals, with the greatest average value in any amnesic individual lower than the lowest average value in any comparison individual (values from the second experiment shown in Fig. P1 C ). Furthermore, on the few occasions when amnesic individuals engaged in spontaneous revisitation, doing so did not enhance their learning, as memory performance for the items studied in this manner was no different from for items studied outside of the setting of revisitation (whereas comparison individuals did show the characteristic performance increase for objects studied with revisitation). To identify mechanisms by which this behavioral optimization was accomplished, we next sought evidence for brain structures and activity associated with spontaneous revisitation. We first examined the behavior of individuals with amnesia caused by severe damage to the hippocampus. The nature of the neural processing accomplished by the hippocampus makes it a crucial structure for long-term memory ( 2 ). The amnesic individuals we studied thus have long-term memory impairments, characterized by relatively brief windows of time during which new information can be maintained before it is forgotten. Memory performance for the specific objects studied during revisitation sequences was significantly more accurate than for the objects that were not studied during revisitation sequences (green vs. black bars, respectively, in Fig. P1 B ). This was found by using two separate test formats: object recognition, involving the discrimination of studied (i.e., old) from unstudied (i.e., new) objects; and spatial recall, involving the attempted replacement of objects to their studied locations. Critically, beneficial effects of revisitation were specific to the volitional study condition, in that there was no difference in accuracy for objects studied with revisitation versus objects studied without revisitation for the passive study condition ( Fig. P1 B ). Memory performance was better overall after volitional study than passive study, as we previously reported ( 1 ), but the effect of revisitation was to enhance memory performance in the volitional condition. Spontaneous revisitation thus represents a form of behavioral optimization because it enhanced learning when it was self-generated, but did not influence learning when the same pattern was merely viewed. As introduced earlier, the phenomenon of interest was the tendency for individuals to spontaneously “look back” in their viewing path to revisit recently seen objects (i.e., view objects in the order A–B–C and then immediately revisit objects in the order B–C as opposed to moving on to object D). This is illustrated by the pattern of object-to-object transitions depicted in Fig. P1 A . On average, 35% of the transitions made from one object to another were part of spontaneous revisitation sequences in a group of healthy, college-age individuals ( n = 34, whose data unrelated to revisitation were originally reported in ref. 1 ). In our experiments, individuals studied sets of object images arranged in a grid pattern on a computer monitor. A “mask” of visual noise occluded the objects, and a small window permitted clear viewing of only one object at a time ( Fig. P1 A ). In the active study condition, individuals used a joystick to move the window to view objects in an entirely self-controlled manner, and this condition was therefore termed “volitional control,” given that individuals could study objects however they chose. This contrasted with the passive condition, in which the window moved under control of the computer and individuals viewed the objects thus revealed to them. The precise visual information provided in the volitional and passive conditions was matched via a subject-to-subject “yoking” procedure, whereby the window movements generated by one individual, n , in the volitional condition were recorded and played back as the passive condition for the next individual, n + 1. Therefore, across individuals, the same visual information was available in both conditions. When animals explore a novel environment, sensory sampling behaviors, such as eye movements, appear continuously optimized so that pertinent information is selected and quickly learned. Little is known of how this seemingly effortless behavioral optimization is accomplished in humans. We therefore examined a form of optimization that occurs spontaneously when individuals study groups of objects: the tendency to “look back” to revisit a subset of recently viewed objects. We found that this spontaneous revisitation improves learning, as indicated by enhanced later memory performance for the specific objects that were revisited. Notably, learning was improved by the act of engaging in revisitation—but not by simply viewing the same objects again—as experiencing the revisitation pattern of another subject in a passive viewing condition did not improve memory. Furthermore, self-generated revisitation was substantially reduced in individuals with amnesia caused by severe damage to the hippocampus, a brain region crucial for long-term memory. In healthy individuals, revisitation correlated with activity in the hippocampus, as well as with areas of prefrontal cortex and cerebellum that are linked to planning and prediction. These findings present a striking example of behavioral optimization in humans and demonstrate the necessity of the hippocampus, along with larger frontocerebellar brain circuitry, in the short-term adaptive control of behavior.
Type of Medium:
Online Resource
ISSN:
0027-8424
,
1091-6490
DOI:
10.1073/pnas.1100225108
Language:
English
Publisher:
Proceedings of the National Academy of Sciences
Publication Date:
2011
detail.hit.zdb_id:
209104-5
detail.hit.zdb_id:
1461794-8
SSG:
11
SSG:
12
Bookmarklink
