The brain network associated with acquiring semantic knowledge

Abstract

There is ongoing debate about how semantic information is acquired, whether this occurs independently of episodic memory, and what role, if any, brain areas such as hippocampus are required to play. We used auditory stimuli and functional MRI (fMRI) to assess brain activations associated with the incidental acquisition of new and true facts about the world of the sort we are exposed to day to day. A control task was included where subjects heard sentences that described novel scenarios involving unfamiliar people, but these did not convey general knowledge. The incidental encoding task was identical for two stimulus types; both shared the same episodic experience (lying in the brain scanner) and conveyed complex information. Despite this, and considering only those stimuli successfully encoded, compared to a baseline task, a more extensive network of brain regions was found to be associated with exposure to new facts including the hippocampus. Direct comparison between the two stimulus types revealed greater activity in dorsal, ventrolateral and dorsomedial prefrontal cortex, medial dorsal nucleus of the thalamus, and temporal cortex for fact stimuli. The findings suggest that successful encoding is not invariably associated with activation of one particular brain network. Rather, activation patterns may depend on the type of materials being acquired, and the different processes they engender when subjects encode. Qualitatively, from postscan debriefing sessions, it emerged that the factual information was found to be potentially more useful. We suggest that current or prospective utility of incoming information may be one factor that influences the processes engaged during encoding and the concomitant neuronal responses.

Introduction

In the course of our lives, we accrue a great deal of general factual knowledge about the world, while also retaining memories of specific events that occurred at a particular time and place. The former is often referred to as semantic memory and the latter as episodic or autobiographical memory (Tulving, 1972). The precise relationship between semantic and episodic memory, particularly when they are initially established, is a matter of debate. It has been argued that the acquisition of semantic knowledge depends on the acquisition of the episode of which it is a part Nadel and Moscovitch, 1997, Nadel and Moscovitch, 1998, Squire and Zola, 1998. Alternatively, it has been proposed that information can be encoded into semantic memory independently of episodic memory, but must be encoded into episodic memory via semantic memory [the serial parallel independent (SPI) model; Tulving, 1995, Tulving, 2001, Tulving and Markowitsch, 1998]. Graham et al. (2000) have questioned the SPI model, suggesting that perceptual information can, in some circumstances, support new episodic learning. Recently, Holdstock et al. (2002) proposed that learning might be better characterized in terms of rapid or slower acquisition irrespective of whether information is semantic or episodic (see also Alvarez and Squire, 1994, Murre, 1996).

In terms of the underlying brain systems, Squire and Zola (1998) believe that there is no compelling evidence to suggest that acquisition of episodic and semantic memory is differentially affected by medial temporal lobe amnesia. Many patients who appear unable to acquire new information, either semantic or episodic (e.g., Cipolotti et al., 2001, Gabrieli et al., 1988, Hamann and Squire, 1995, Shimamura and Squire, 1987, Verfaellie et al., 1995), are reported. Mishkin et al., 1997, Mishkin et al., 1998, however, proposed that the hippocampus is necessary for the acquisition of new episodic memories, while semantic learning can be supported by subhippocampal cortices. Patients have been reported to show some preservation of the ability to acquire semantic information in the context of impaired episodic acquisition (e.g., Baddeley et al., 2001, Glisky et al., 1986, Kitchener et al., 1998, McKenna and Gerhand, 2002, Tulving et al., 1991, Van der Linden et al., 1994, Van der Linden et al., 1996, Van der Linden et al., 2001, Vargha-Khadem et al., 1997, Verfaellie et al., 2000). The patients described by Vargha-Khadem et al. (1997) offer particularly striking evidence to suggest the independence of semantic and episodic memory, and the necessity for the hippocampus only for the latter.

These cases of developmental amnesia, with apparently selective bilateral hippocampal pathology, were impaired at retaining new episodic memories, but they had relatively intact general factual knowledge about the world and were able to study in mainstream education. However, a patient with adult-acquired selective bilateral hippocampal pathology (YR) was reported to have a significant memory impairment for factual information to which there had been limited exposure and a less severe deficit for factual information to which she had been exposed repeatedly (Holdstock et al., 2002). It was concluded that the hippocampus is critical for rapidly learned information, and while probably also making a contribution to slow learning, this is primarily mediated by the neocortex. Holdstock et al. further predicted that repeated exposure to a personally experienced episode would result in it eventually being learned even in the absence of a functioning hippocampus. Nadel and Moscovitch (1997) also consider the hippocampus to be critical for the rapid learning of both episodic and semantic information, but in contrast to Holdstock et al. (2002), they believe that episodic memories are always dependent on the hippocampus.

In summary, one prevailing view is that the hippocampus is involved in the initial acquisition of both semantic and episodic memories, either because semantic information is not initially separable from an episodic experience Nadel and Moscovitch, 1997, Nadel and Moscovitch, 1998, Squire and Zola, 1998 or because of the hippocampal role in rapid acquisition in general (Holdstock et al., 2002). The proposed hierarchical hippocampal memory system described by Mishkin et al., 1997, Mishkin et al., 1998 predicts the involvement of some or all of the subhippocampal (parahippocampal, perirhinal, and entorhinal) cortices during semantic acquisition. Given the memory profiles of the developmental amnesic cases, it may be that the hippocampus is not involved in semantic acquisition. Alternatively, in healthy brains, all incoming information may be encoded into episodic memory, although this is not necessary for semantic acquisition to occur (Tulving et al., 1991).

Is there converging evidence for any of these predictions from neuroimaging studies involving healthy subjects? There is a large literature of neuroimaging studies investigating the encoding of items novel to the experimental context, such as words, pictures, and pairs of items. However, we were specifically interested in the acquisition of new real-world facts. By this, we mean the kinds of general information we are exposed to day to day, rather than decontextualized single or paired items. As far as we are aware, the acquisition of new and real facts has not been examined using neuroimaging. Therefore, to throw further light on how we accumulate knowledge crucial to our semantic memory store, we set out to investigate in healthy subjects the brain areas associated with the initial and incidental hearing of novel factual information.

A control task was included where subjects also heard sentences, but these did not convey general knowledge (see Appendix A for examples). The control task had several purposes. It controlled for basic auditory, lexical, semantic, attentional, and motor factors, as well as novelty, as sentences in both tasks were novel. Subjects had no a priori knowledge of the different sentences types; the stimuli were randomly interleaved in an event-related design, and there was no explicit instruction to learn. Acquisition of semantic information occurs initially in a particular context. In this case, however, the primary effect of the episodic experience of lying in the scanner was the same for both the experimental and control stimuli.

If, as the prevailing views above suggest, incoming information is not initially separable from an episodic experience, or if all incoming information is encoded into episodic memory although this is not necessary for semantic acquisition to occur, then there should be no difference between experimental and control tasks in terms of underlying activation patterns. Alternatively, if semantic information can be encoded without the need for the hippocampus, there may be attenuation of hippocampal activity during fact acquisition.