HIPPOCAMPUS 20:889–893 (2010)

RAPID COMMUNICATION Long-Term Visuospatial Retention Unaffected by Fornix Transection Sze Chai Kwok* and Mark J. Buckley ABSTRACT: As part of an earlier experiment (Kwok and Buckley, 2009), six macaque monkeys (three with fornix transection and three unoperated controls) were trained postoperatively to discriminate a total of 104 new concurrent visuospatial conditional problems to criterion. Our experiment measured and compared long-term retention of these problems with two separate one-trial postoperative retention tests administered 3 and 15 months, respectively, after acquisition. All animals showed some degree of forgetting of these problems but all remembered above chance levels, even after 15 months. The amount forgotten by each group did not differ significantly at either time point. These results show that long-term retention of visuospatial information is independent of the fornix. Similarities in resistance to forgetting are drawn between fornix-transected macaques and patients with amnesia and the implications for clinical rehabilitation are discussed. V 2009 C

Wiley-Liss, Inc.

KEY WORDS: rehabilitation

forgetting; macaques; hippocampus; memory; amnesia;

In humans, damage to the fornix is known to impair anterograde memory (e.g., Gaffan and Gaffan, 1991; Gaffan et al., 1991; Hodges and Carpenter, 1991; McMackin et al., 1995; Aggleton et al., 2000; Vann et al., 2000; Papanicolaou et al., 2007), but fornix damage has also been linked to remote memory deficits (Spiers et al., 2001), retrograde temporal order amnesia (Yasuno et al., 1999; Squire et al., 2001), and retrograde amnesia (Poreh et al., 2006). The fornix’s intimate anatomical connection with the mammillary bodies of the medial diencephalon (Saunders and Aggleton, 2007)—a structure that has itself been implicated in retrograde memory (Squire et al., 1989; Kopelman, 2002)—also suggests that the fornix might be involved in the retention of previously acquired memories. On the other hand, in macaque monkeys in which highly selective and complete fornix transection can be reliably induced by surgical intervention, it has been established that fornix transection affects new postoperative learning (i.e., anterograde amnesia) far more than it affects retrieval of previously acquired memories (i.e., retrograde amnesia). For example, earlier studies that looked at the effects of fornix transection upon retention of preoperatively acquired discriminations within algorithmically generated complex scenes (see Gaffan, 1994, task 1), or of postoperatively learned discriminations between complex naturalistic scenes (Gaffan, 1993), found that while Department of Experimental Psychology, University of Oxford, Oxford, United Kingdom Grant sponsors: MRC Program, China Oxford Scholarship Fund. *Correspondence to: Sze Chai Kwok, Department of Experimental Psychology, South Parks Road, Oxford OX1 3UD, United Kingdom. E-mail: [email protected] Accepted for publication 28 September 2009 DOI 10.1002/hipo.20733 Published online 15 December 2009 in Wiley InterScience (www.interscience. wiley.com). C 2009 V

WILEY-LISS, INC.

new learning was impeded, retention of these discriminations was preserved. In Gaffan’s (1993) study, unoperated control and fornix-transected monkeys were trained postoperatively to learn to discriminate among 320 complex naturalistic scenes and were subsequently tested on their extent of forgetting of these scenes after reaching criterion. Despite a reduced rate of acquisition of the fornixtransected monkeys, the number of forgotten scenes was equal in the two groups of monkeys on a retention test at 49 days later. However, neither the complex naturalistic scene task (Gaffan, 1993) nor the algorithmically generated complex scene task (see Gaffan, 1994, task 1) necessitated spatial memory per se (as memories for foreground objects or background detail can aid the performance of each) and so it is not possible to infer from these studies that retrieval of spatial memory is unaffected by fornix transection. For this reason, Buckley et al. (2004) assessed the effects of fornix transection on a concurrent spatial discrimination learning task to provide a direct assessment of spatial memory; further, an one-trial postoperative retention test paradigm was used to allow the effects of the lesion upon retention to be quantified independently from effects upon relearning. Buckley et al. (2004) found that immediate postoperative retention of 40 preoperatively learned visuospatial problems was unaffected by fornix transection, but subsequent new postoperative learning of similar problems was significantly impaired. Similarly, using a modified concurrent spatial discrimination paradigm, Buckley et al. (2008) confirmed that immediate postoperative recall of 288 preoperatively learned visuospatial problems was likewise unaffected by fornix transection, whereas new postoperative visuospatial learning was again impaired. Nevertheless, the one-trial postoperative retention tests in these two studies [and also in the study by Gaffan (1994)] were administered within a relatively short period of time following visuospatial memory acquisition and subsequent surgery (2 weeks on average). Thus, these studies, while providing robust evidence that fornix transection does not affect short-term retention, do not speak toward the effect of fornix transection on long-term retention. And Gaffan’s (1993) study, which employed a relatively longer period of 49 days before a subsequent retention test was given, did not necessitate the use of spatial memory as highlighted above. Therefore, to

890

KWOK AND BUCKLEY

FIGURE 1. The two panels show two illustrative examples of problems from the visuospatial conditional discrimination. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

facilitate a direct assessment of the effect of time on the longterm retention of visuospatial memories after fornix transection, we taught groups of fornix-transected and control monkeys a total of 104 new concurrent visuospatial problems to criterion, postoperatively, and then gave them two one-trial postoperative retention tests which were administered after much longer periods of time had elapsed after acquisition, namely, 3 and 15 months, respectively. The subjects were six male cynomolgus monkeys (Macaca fascicularis), three of which were unoperated controls and three of which had received fornix transection. Their mean weight at the start of this study was 6.1 kg (range, 5.2–7.0 kg), and their mean age was 4 years and 9 months. All licensed procedures were carried out in compliance with the United Kingdom Animals (Scientific Procedures) Act of 1986. Microscopic examination of the stained histology sections confirmed that in all three lesioned cases there was a complete section of the fornix with no damage outside the fornix except for the incision in the corpus callosum and, at most, only slight damage to the most ventral part of the cingulate gyrus at the same anterior–posterior level in only one hemisphere of one animal. Detailed photomicrographs of the histology for the lesioned animals together with descriptions of the surgical method have been presented elsewhere (Buckley et al., 2008). The behavioral testing was performed in an automated testing cubicle described in detail elsewhere (Buckley et al., 2008), which consisted, briefly, of a touch-sensitive screen (380 mm wide and 280 mm high) on which visual stimuli could be displayed and which could be touched by the subject from its transport cage to obtain food reward pellets (190 mg; P.J. Noyes, Lancaster, NH) consequent upon correct choices. The animal’s daily diet of wet monkey chow, pieces of fruits, raisins, and peanuts was provided in an automated lunchbox that opened once the session was completed. The animals had on average of 10 months of experience in a visuospatial concurrent discrimination learning task (with two-choice problems) that did not form part of our experiment and that has been reported separately (Buckley et al., 2008). During the course of Hippocampus

that study the six animals were divided into two performancematched groups, one group (CON group) remained unoperated controls and the other received bilateral transection of the fornix (FNX group). After that study was completed the six animals were commenced further for postoperative testing involving acquisition of a further 104 new postoperative visuospatial conditional problems of a different kind (i.e., the fourchoice problems of the our study) (Kwok and Buckley, 2009). Kwok and Buckley (2009) reported that although the FNX group was unimpaired in terms of numbers of error accrued while learning large sets of visuospatial problems to criterion (the CON group accumulated a mean of 1,640 errors in learning the 104 problems, and the corresponding mean was 1,995 errors for the FNX group), the FNX group was impaired during the rapid acquisition phases of the visuospatial associations as was apparent from their impaired ability to learn the smallest set as quickly as controls and their reduced ability to eliminate errors in the earliest phases of learning sets of any size. Our study concerns our analyses of the long-term forgetting of these 104 problems that took place 3 and 15 months after acquisition, respectively. In between these two retention tests, the monkeys were also trained to discriminate randomly changing pairings in an object discrimination learning task (Wilson et al., 2007). The visuospatial problems in this experiment consisted of four identical multicolored cartoon-like stimuli (each subtending a visual angle of 11.58) presented upon a white background, and the four identical stimuli in each problem were presented in four fixed positions, symmetrically, with two at the top and two at the bottom on the touch screen (see Fig. 1). For each problem, one out of the four positions in which the four copies of the identical stimulus appeared was predesignated as the rewarded position (S1) and the other three positions were unrewarded foils; the position–reward contingency for each particular problem remained constant throughout the experiment and there were equal numbers of problems in which each of the four positions were rewarded. Thus, conditional upon a problem’s object identity, the mon-

UNIMPAIRED RETENTION AFTER FORNIX TRANSECTION

FIGURE 2. The level of retention of 104 visuospatial concurrent discrimination problems in two retention tests, RET3M and RET15M, administered 3 and 15 months, respectively, after postoperative learning to a 90% performance criterion occurred.

keys learned, by trial and error, which one of the four possible positions was the choice that each problem instructed. As described in the study by Kwok and Buckley (2009), a correction trial procedure was adopted during acquisition of the 104 problems and the problems were acquired concurrently in three consecutive sets (of 8, 32, and 64 problems, respectively) to a criterion level of performance of 90% correct in each case. The requirement of a 90% criterion performance level of these 104 problems in all animals augmented the credibility of the ensuing retention tests. In our study the monkeys were given two retention tests (RET3M and RET15M administered 3 and 15 months after learning) in which each of the 104 problems which they had previously acquired was presented once in a single session in a random order. On the day prior to each of the retention tests, a short refamiliarization test was given to remind the monkeys about the testing conditions and to ensure that the monkeys had not altered their motivation as a consequence of the extended breaks from testing. The refamiliarization test comprised 12 different problems (taken from a preliminary training stage) with each of these problems presented three times in this single session. The retention rates of these 104 problems in RET3M were 54.2% for the CON group and 55.8% for the FNX group (Fig. 2). The retention rates of these 104 problems in RET15M were 45.5% for the CON group and 49.4% for the FNX group (Fig. 2). Paired-sample t-tests showed significant forgetting in both groups after 3 months [RET3M: CON: t(2) 5 9.19, P 5 0.012; FNX: t(2) 5 8.55, P 5 0.013] and after 15 months [RET15M: CON: t(2) 5 9.58, P 5 0.011; FNX: t(2) 5 11.39, P 5 0.008] compared with the 90% performance levels attained after acquisition. To ascertain whether forgetting proceeded at the same rate in the two groups, we

891

ran a two-way repeated measures analysis of variance with two levels of a between-subjects factor ‘‘Group’’ (CON, FNX) and two levels of a within-subjects factor ‘‘Time’’ (RET3M, RET15M) on the percentage of correct responses from the retention tests. We found neither a main effect of ‘‘Group’’ [Group: F (1,4) < 1] nor a main effect of ‘‘Time’’ [Time: F (1,4) 5 2.94, P > 0.1]. The ‘‘Group 3 Time interaction’’ was also insignificant [Group 3 Time: F (1,4) < 1]. Thus the groups did not differ in their forgetting rates. Despite the forgetting exhibited by both groups of monkeys over a period of 15 months, the animals still remembered above a chance level of 25% (as our task was a series of independent four-choice problems) after 3 months [RET3M: CON: t(2) 5 7.48, P 5 0.017; FNX: t(2) 5 7.69, P 5 0.017] and after 15 months [RET15M: CON: t(2) 5 4.43, P 5 0.047; FNX: t(2) 5 6.83, P 5 0.021]. To test whether animals retained the same specific problems as each other as might be expected if some problems were more distinct and easier to remember than others, we ran a series of pairwise comparisons between individual animals within groups to assess what degree of correspondence there might be between individuals with respect to which specific problems were remembered; this was done for both retention tests separately. The mean group correlation coefficients on RET3M and RET15M for the FNX group were 0.127 and 0.161; the corresponding coefficients for the CON group were 0.199 and 0.108. All but two within-group correlations between individuals were insignificant. There were also no group differences (all P > 0.5), suggesting that there was no correspondence between which particular problems individual monkeys remembered within, or between, groups. Moreover, we ascertained the proportion of problems correctly remembered at the first retention test which were subsequently also retained in the second retention tests: In RET15M, the CON group remembered a mean of 58% of the problems that were retained correctly on RET3M, the corresponding mean for the FNX group was 60%. Specific problems remembered on RET3M were significantly more likely retained on RET15M than a chance level (25%) in all monkeys (P < 0.01), and there were also no group differences in this respect (P > 0.5). Taken together, these analyses suggest that, while all monkeys demonstrated significant forgetting, the FNX group exhibited a forgetting rate indistinguishable from that of CON monkeys. We conclude that long-term retention of these visuospatial problems is not fornix dependent. Our study thus extends our knowledge that long-term forgetting is not affected by fornix transection from a period as long as 49 days (Gaffan, 1993) to a much longer period, i.e., up to 15 months. In addition, we showed that fornix-transected monkeys, while clearly being unimpaired compared with control monkeys in long-term retention memory, still remembered the problems significantly above chance even after 15 months. Some components of their visuospatial memory traces are therefore preserved for at least 15 months (extrapolation of forgetting rates would suggest for much longer periods than this) in a manner that is independent of the fornix. Hippocampus

892

KWOK AND BUCKLEY

Although our task is a visuospatial conditional task in which a certain visual stimulus instructs which one of the four positions will be rewarded, it is likely that learning of the current task, in normal animals, may proceed by a combination of fornix-dependent and fornix-independent learning strategies. Previous studies have shown that the fornix is important for discriminating between scenes that vary in the spatial arrangements of features (Gaffan and Harrison, 1989b) and for learning about the spatial organization of complex scenes (Gaffan, 1991); moreover, the fornix is necessary for spatial configural learning (see Buckley et al., 2004, exp. 3) and spatial conditional learning paradigms (Gaffan and Harrison, 1989a). Although the role of the fornix in spatial learning is well established, it is equally apparent that the fornix is not necessary for all forms of spatial learning. Discriminating among locations in space and associating those locations with objects is independent upon the fornix (Murray et al., 1989), as is place discrimination if the scenes are sufficiently distinct (Gaffan and Harrison, 1989b). In our task, the four identical items on the screen could be described as distinct ‘‘scenes’’ that are unique to each problem. Moreover, the use of a correction procedure in the acquisition of the 104 visuospatial problems might arguably promote some degree of successive learning, and Gaffan et al. (1984, exp. 6) found that spatial conditional discriminations were unimpaired after fornix transection, showing that they were successively and not concurrently learned. We speculate that fornix-transected monkeys can employ spatial learning strategies that are independent of the fornix, particularly when they are faced with tasks of a familiar format that do not place high demands on contextual and configural learning. This hypothesis is consistent with Gaffan’s (1993) suggestion that the fornix-transected animals learned differently from control animals, relying more on local object cues and less on whole-scene spatial cues than the normal animals. Nevertheless, our study shows that even if CON monkeys do have additional strategies for learning compared with FNX monkeys, both groups forget at the same rate, and hence, different routes to learning do not dictate how things are subsequently remembered. It is established that longer retention intervals lead to more forgetting (King et al., 2002), and forgetting functions typically decrease with time, as is evident with our findings too. At least three candidate factors have been proposed to explain forgetting in animals, most commonly proactive interference (PI), retroactive interference (RI), and memory trace deterioration (memory decay) (for a review, see Wixted, 2004). For example, it is known that prior learning can profoundly affect the forgetting of subsequently learned material (Underwood, 1957), particularly when the prior learning trials were massed (Underwood and Ekstrand, 1966, 1967). Ample evidence also points to a theory of forgetting processes associated with the formation of new memories which retroactively interfere with previously formed memories that are still undergoing the process of consolidation (Villarreal et al., 2002; Wixted, 2004). With respect to the our study, the monkeys were trained on tasks such as concurrent spatial discrimination learning (Buckley et al., 2008) prior to commencement of our study and were exposed Hippocampus

to tasks including object discrimination learning with randomly changing pairings (Wilson et al., 2007) before the last retention test. According to the interference theories, these pre- and postexposure to various kinds of learning problems and scenarios should lead to both PIs and RIs to the learned material in the current task, and so the extent to which each of these interfering factors contributes is difficult to determine. One also cannot rule out that time-dependent ‘‘trace deterioration’’ itself may also play a significant role in forgetting (Bailey and Chen, 1989). Finally, with a couple of points of qualification, our data may speak toward the potential for amnesic patient rehabilitation. The first point of qualification is whether patients are in general as extensively exposed to information during acquisition as our monkeys were; however, because we ceased acquisition training of our animals as soon as they attained a criterion of 90% we would suggest that there is a relatively slim possibility that the lack of lesion-mediated effect on long-term retention in our study is largely attributable to extensive training of problems prior to the measure of long-term retention. A second point of qualification concerns the extent to which we should generalize our findings beyond conditional stimulus–reward learning tasks of the kind used here to other forms of nondeclarative memory, or even some forms of declarative information. Although it is certainly beyond the scope of our study to determine whether information acquired by our animals is in any way available to conscious inspection (see Tulving, 2002; Clayton et al., 2003 for opposing views on this debate), it is noteworthy that some studies have shown that even animals exhibit behavior that is suggestive of processes akin to conscious recollection existing at some level, even in rodents (e.g., Fortin et al., 2004). Regarding this, our data lend support to the idea that if appropriate techniques such as rehearsal and errorless learning (Tailby and Haslam, 2003; Clare and Jones, 2008) are utilized to assist amnesic patients to learn certain associations or reacquire what was forgotten (Miotto, 2007), then such patients may, under some circumstances, be able to retain what has been learned at levels similar to the retention abilities of normal healthy subjects and for extended periods of time.

Acknowledgment The authors thank David Gaffan for continued support while conducting this research project.

REFERENCES Aggleton JP, McMackin D, Carpenter K, Hornak J, Kapur N, Halpin S, Wiles CM, Kamel H, Brennan P, Carton S, Gaffan D. 2000. Differential cognitive effects of colloid cysts in the third ventricle that spare or compromise the fornix. Brain 123:800–815. Bailey CH, Chen M. 1989. Time course of structural changes at identified sensory neuron synapses during long-term sensitization in aplysia. J Neurosci 9:1774–1780.

UNIMPAIRED RETENTION AFTER FORNIX TRANSECTION Buckley MJ, Charles DP, Browning PGF, Gaffan D. 2004. Learning and retrieval of concurrently presented spatial discrimination tasks: Role of the fornix. Behav Neurosci 118:138–149. Buckley MJ, Wilson CRE, Gaffan D. 2008. Fornix transection impairs visuospatial memory acquisition more than retrieval. Behav Neurosci 122:44–53. Clare L, Jones RSP. 2008. Errorless learning in the rehabilitation of memory impairment: A critical review. Neuropsychol Rev 18:1–23. Clayton NS, Bussey TJ, Dickinson A. 2003. Can animals recall the past and plan for the future? Nat Rev Neurosci 4:685–691. Fortin NJ, Wright SP, Eichenbaum H. 2004. Recollection-like memory retrieval in rats is dependent on the hippocampus. Nature 431: 188–191. Gaffan D. 1991. Spatial organization of episodic memory. Hippocampus 1:262–264. Gaffan D. 1993. Normal forgetting, impaired acquisition in memory for complex naturalistic scenes by fornix-transected monkeys. Neuropsychologia 31:403–406. Gaffan D. 1994. Scene-specific memory for objects: A model of episodic memory impairment in monkeys with fornix transection. J Cogn Neurosci 6:305–320. Gaffan D, Gaffan EA. 1991. Amnesia in man following transection of the fornix: A review. Brain 114:2611–2618. Gaffan D, Harrison S. 1989a. A comparison of the effects of fornix transection and sulcus principalis ablation upon spatial learning by monkeys. Behav Brain Res 31:207–220. Gaffan D, Harrison S. 1989b. Place memory and scene memory: Effects of fornix transection in the monkey. Exp Brain Res 74: 202–212. Gaffan D, Saunders RC, Gaffan EA, Harrison S, Shields C, Owen MJ. 1984. Effects of fornix transection upon associative memory in monkeys: Role of the hippocampus in learned action. Q J Exp Psychol: Comp Physiol Psychol 36:1173–1221. Gaffan EA, Gaffan D, Hodges JR. 1991. Amnesia following damage to the left fornix and to other sites: A comparative study. Brain 114:1297–1313. Hodges JR, Carpenter K. 1991. Anterograde amnesia with fornix damage following removal of IIIrd ventricle colloid cyst. J Neurol Neurosurg Psychiatry 54:633–638. King DL, Jones FL, Pearlman RC, Tishman A, Felix CA. 2002. The length of the retention interval, forgetting, and subjective similarity. J Exp Psychol: Learn Memory Cogn 28:660–671. Kopelman MD. 2002. Disorders of memory. Brain 125:2152–2190. Kwok SC, Buckley MJ. 2009. Fornix transection selectively impairs fast learning of conditional visuospatial discriminations. Hippocampus 10.1002/hipo.20643. McMackin D, Cockburn J, Anslow P, Gaffan D. 1995. Correlation of fornix damage with memory impairment in six cases of colloid cyst removal. Acta Neurochir 135:12–18. Miotto EC. 2007. Cognitive rehabilitation of amnesia after virus encephalitis: A case report. Neuropsychol Rehabil 17:551–566.

893

Murray EA, Davidson M, Gaffan D, Olton DS, Suomi S. 1989. Effects of fornix transection and cingulate cortical ablation on spatial memory in rhesus monkeys. Exp Brain Res 74:173– 186. Papanicolaou AC, Hasan KM, Boake C, Eluvathingal TJ, Kramer L. 2007. Disruption of limbic pathways in a case of profound amnesia. Neurocase 13:226–228. Poreh A, Winocur G, Moscovitch M, Backon M, Goshen E, Ram Z, Feldman Z. 2006. Anterograde and retrograde amnesia in a person with bilateral fornix lesions following removal of a colloid cyst. Neuropsychologia 44:2241–2248. Saunders RC, Aggleton JP. 2007. Origin and topography of fibers contributing to the fornix in macaque monkeys. Hippocampus 17: 396–411. Spiers HJ, Maguire EA, Burgess N. 2001. Hippocampal amnesia. Neurocase 7:357–382. Squire LR, Haist F, Shimamura AP. 1989. The neurology of memory: Quantitative assessment of retrograde amnesia in two groups of amnesic patients. J Neurosci 9:828–839. Squire LR, Clark RE, Knowlton BJ. 2001. Retrograde amnesia. Hippocampus 11:50–55. Tailby R, Haslam C. 2003. An investigation of errorless learning in memory-impaired patients: Improving the technique and clarifying theory. Neuropsychologia 41:1230–1240. Tulving E. 2002. Episodic memory: From mind to brain. Annu Rev Psychol 53:1–25. Underwood BJ. 1957. Interference and forgetting. Psychol Rev 64: 49–60. Underwood BJ, Ekstrand BR. 1966. An analysis of some shortcomings in the interference theory of forgetting. Psychol Rev 73: 540–549. Underwood BJ, Ekstrand BR. 1967. Studies of distributed practice: XXIV. Differentiation and proactive inhibition. J Exp Psychol Appl 74:574–580. Vann SD, Brown MW, Erichsen JT, Aggleton JP. 2000. Using Fos imaging in the rat to reveal the anatomical extent of the disruptive effects of fornix lesions. J Neurosci 20:8144–8152. Villarreal DM, Do V, Haddad E, Derrick BE. 2002. NMDA receptor antagonists sustain LTP, spatial memory: Active processes mediate LTP decay. Nat Neurosci 5:48–52. Wilson CRE, Charles DP, Buckley MJ, Gaffan D. 2007. Fornix transection impairs learning of randomly changing object discrimination. J Neurosci 27:12868–12873. Wixted JT. 2004. The psychology and neuroscience of forgetting. Annu Rev Psychol 55:235–269. Yasuno F, Hirata M, Takimoto H, Taniguchi M, Nakagawa Y, Ikejiri Y, Nishikawa T, Shinozaki K, Tanabe H, Sugita Y, Takeda M. 1999. Retrograde temporal order amnesia resulting from damage to the fornix. J Neurol Neurosurg Psychiatry 67:102–105.

Hippocampus