Inter-relationships between attention, activation, fMR adaptation and long-term memory

Abstract

fMR adaptation in the ventral visual pathway reflects information processing that may contribute to implicit and explicit memory. In experiments that employed < 1 s repetition lag, we found that attention increases adaptation for repeated objects in brain regions at the top of the visual processing hierarchy (anterior fusiform and parahippocampal gyri) but that it can still appear with minimal attention in most of the fusiform bilaterally. Of the ventral visual regions showing adaptation, the parahippocampal region and LOC showed the strongest correlation between adaptation magnitude and recognition memory across subjects. Although there was some overlap, regions showing correlations between adaptation and priming lay more posteriorly within the fusiform region. The positive association between encoding-related activation and adaptation suggests that over an entire test set, memory performance can be determined by neural events occurring in the peristimulus period. This may reflect stronger engagement of attention at encoding.

Introduction

Adaptation, also termed repetition suppression (Grill-Spector and Malach, 2001, Henson and Rugg, 2003), is thought to reflect stimulus specific perceptual memory and its magnitude has been used to index item resemblance (Grill-Spector and Malach, 2001, Grill-Spector, 2004) as well as memory strength (Epstein et al., 2005, Turk-Browne et al., 2006).

Along the ventral visual stream, repetition induced neural response attenuation has been shown to be greater when a stimulus is explicitly remembered (Turk-Browne et al., 2006). Interestingly, adaptation can also be found when behavioral differences between initial and repeat presentations are absent (Maccotta and Buckner, 2004, Klaver et al., 2007) and when test stimuli are not consciously perceived (Dehaene et al., 2001). In addition, it has been observed at locations differing in functional specialization (Grill-Spector et al., 2006), suggesting that the neural mechanisms underlying adaptation vary according to experimental context and neuroanatomical locale (Schacter et al., 2007). For instance, it has been demonstrated for repeated contours in early visual areas (Kourtzi and Huberle, 2005), for repeated shapes in lateral occipital cortex (Murray and Wojciulik, 2004) and in the parahippocampal gyrus for repeated black and white scenes (Yi and Chun, 2005).

In this work, we clarify the relationships among attention, encoding-related activation, adaptation and long-term explicit memory (henceforth, the term ‘activation’ refers to ‘encoding-related activation’). Specifically, our goal was to evaluate how attention may modulate adaptation at various points along the ventral visual pathway noting that links between repetition effects and the prefrontal cortex are already well established (Dobbins et al., 2004, Maccotta and Buckner, 2004, Wig et al., 2005). We also sought to uncover where along this pathway adaptation and long-term explicit memory correlate best (Turk-Browne et al., 2006), thus differentiating the present work from the many adaptation studies of implicit memory/priming.

Attention during encoding strongly influences later memory performance (Mack and Rock, 1998). For example, paying attention to a stimulus benefits memory whereas dividing attention impairs performance on many measures of memory (Craik et al., 1996, Mulligan, 1998). The neural counterpart to this behavioral finding is that an attended stimulus is accompanied by relatively greater activation of the ventral visual cortex than an unattended stimulus (Brefczynski and DeYoe, 1999, Gandhi et al., 1999, Kastner and Ungerleider, 2000). Further, activation in stimulus specific cortex is relatively suppressed for ignored items (Gazzaley et al., 2005). Finally, across many studies, there is a strong and consistent relationship between activation magnitude at encoding in a number of brain regions and subsequent memory (Brewer et al., 1998, Wagner et al., 1998, Kirchhoff et al., 2000).

The relationship between attention and adaptation is more complex. On the one hand, attention has been shown to be critical for adaptation (Eger et al., 2004, Murray and Wojciulik, 2004, Vuilleumier et al., 2005, Yi and Chun, 2005). On the other hand, stimulus repetition effects have also been observed without conscious awareness (Schott et al., 2005), suggesting that adaptation can be automatic (Wiggs and Martin, 1998), depending on the task and on where in the cortex one looks. Furthermore, the correlation between the magnitude of adaptation and memory appears to be task dependent, there being reports of both clear (Epstein et al., 2005, Turk-Browne et al., 2006) and absent (Maccotta and Buckner, 2004) correlations between adaptation within the ventral visual pathway and memory.

To reconcile some of the inconsistent past findings, we posit that a part of the adaptation response within the ventral visual pathway may arise from visual processing sensitive to perceptual repetition and may occur with minimal engagement of attention. This component would not be expected to contribute to the formation of an explicitly remembered percept, unlike the attention-sensitive component that contributes to explicit memory (see Slotnick and Schacter, 2004). A similar dissociation between ventral visual regions showing adaptation with and without accompanying behavioral priming has been recently uncovered (Ganel et al., 2006, Sayres and Grill-Spector, 2006).

Many recent experiments evaluating the importance of attention in adaptation used paradigms that require ignoring the stimulus that would otherwise generate adaptation (Eger et al., 2004, Murray and Wojciulik, 2004, Vuilleumier et al., 2005, Yi and Chun, 2005). These paradigms studied the impact of divided attention on adaptation using strong manipulations of attention, whereas the original descriptions of repetition suppression in primate work referred to experiments involving passive viewing, where attention was not necessarily engaged (Baylis and Rolls, 1987, Fahy et al., 1993, Li et al., 1993). The current work seeks to study the effect of attention somewhere in the middle of the continuum (Klaver et al., 2007) given that the manner in which attention is divided influences adaptation and memory (Chun and Turk-Browne, 2007).

Another related issue pertains to the link between memory and where adaptation is observed. Although many areas in the ventral visual cortex demonstrate adaptation, adaptation may be correlated with explicit memory only in a subset of these regions (Turk-Browne et al., 2006). Also, while behavioral priming and repetition suppression have been shown to occur at the same time, they may not be correlated (Maccotta and Buckner, 2004, McMahon and Olson, 2007). Here, we sought to show how adaptation in different regions might contribute to implicit and explicit memory.

Finally, behavioral (Bentin and Moscovitch, 1988), electrophysiological (Nagy and Rugg, 1989) and functional imaging (Henson et al., 2004) work suggests a distinction between short and long-lag repetition effects. In terms of pure repetition suppression magnitude, short lags between repeated and novel stimuli generate significantly larger effects than longer lags (Baylis and Rolls, 1987, McMahon and Olson, 2007). Given that the magnitude of adaptation with long-lag repetition in the PPA tracks long-term explicit memory (Turk-Browne et al., 2006), we would expect that short-lag (< 1 s.) repetition effects should be as, if not more, predictive of such memories (Brozinsky et al., 2005). Short-lag repetition designs also serve to eliminate interference effects from intervening stimuli arising from semantic similarity (see Klaver et al., 2007). Highly emotional stimuli might also interfere with subsequent retrieval. An extreme example of this is how the interposition of emotional stimuli can disrupt the maintenance of visual memories (Dolcos and McCarthy, 2006).