When being asked about the importance of memory most people would say that remembering something is good and forgetting something is bad. While this seems reasonable to believe about factual knowledge (f.e. information about places or names of persons), there are certainly cases where it is beneficial to not remember something. The most prominent example is amnesia after a traumatic event. But forgetting might not only be useful in extreme situations but also in everyday experiences. For example, if you lost some money on your way to work you would prefer to not have this thought flash into your awareness all the time and rather forget about it. This means you actively try to purge those memories from your conscious awareness.

Recent scientific studies have given us a good insight in how conscious experiences may be represented as neural substrates. However, we still have only limited knowledge about how overt control of the content of our conscious experiences can be achieved. Is it possible to consciously purge memories from our awareness if they are undesired? In his presentation, Michael Anderson presents evidence that this strategic manipulation of memories in conscious awareness is indeed possible and directly influences what we do and do not remember.

One way to study these processes is behaviorally investigating whether it is possible to willingly stop memory retrieval. For this purpose, Anderson and Green (2001) developed the Think/No-Think paradigm which is based on the prominent Go/No-Go paradigm. Participants were instructed to learn associations between two words representing objects, places or similar. Afterwards, they performed the crucial Think/No-Think task consisting of two types of trials: respond trials and suppression trials. In both types of trials they were presented a word. In respond trials they were asked to actively bring the associated word into consciousness while they should avoid thinking of the associated word in suppression trials. Using a follow-up recall task 32 studies using this paradigm revealed that retrieval of suppressed words was significantly less likely than retrieval of responded words and also less likely than the baseline condition. This implies that reminding cues can not only strengthen retrieval but also suppress it depending on your response to the cue.

Hippocampal modulation and systemic control

After establishing that attentional control processes for memory suppression indeed exist, it is of scientific interest to find the relating neural mechanisms. In a study employing functional magnet resonance imaging (fMRI) Anderson et al. (2004) investigated what happens in the brain when we try to control undesired memories. Since many prior studies suggested that the prefrontal cortex plays the decisive role in attentional control (see review by Miller, 2000) it was the first investigated brain region. They were able to show that dorsolateral prefrontal cortex (DLPFC) activity directly correlated with successful memory recall. When DLPFC activity was in the 1^st quartile ~90% of the words were recalled correctly as opposed to ~75% correct retrievals when DLPFC activity was in the 4^th quartile. Besides, they showed that hippocampal activation is reduced when memories are suppressed. The result was interpreted as an indication that the activation in the hippocampus during memory retrieval can be modulated by cognitive control. This led to two hypotheses:

Hippocampal Modulation: Hippocampal activation can be actively altered to purge episodic memories from awareness.

Systemic Control: Hippocampal activation is globally modulated by cognitive control.

The study revealed correlations between activation in brain regions and percentage of correctly recalled memories. However, strictly speaking it could not make an inference about purging of memories because it neglected the possibility that a memory episode has never reached conscious awareness at all. For purging to have occurred, the memory episode must have been in conscious awareness and then voluntarily pushed out of it. Therefore, it needs to be ensured that the memory episode was in conscious awareness at some point.

Levy and Anderson (2012) approached this problem by asking participants how often the to-be-suppressed memories flashed into their consciousness (as intrusions) in the No-Think condition. The fMRI data analysis led to the same results which means that the added question did not interfere with the control processes in question. The results showed a strong correlation (r = .70) between intrusion activity and signal reduction in the hippocampus while there was no correlation for non-intrusion activity. This suggests that intrusions trigger the respective control mechanisms for memory suppression and purging indeed occurs as the memory episodes entered conscious awareness. However, the authors acknowledged that one has to be careful to draw conclusions from these results alone. It remained unclear whether the reduced hippocampal activity is really due to direct suppression of memories. An explanation could also be that the participants did not directly suppress the memory but rather substituted it with another low-level thought.

Mechanisms of mnemonic control and the amnesic penumbra

Benoit and Anderson (2012) claim there are two distinct processes of mnemonic control: direct suppression and thought substitution. In their study they tried to dissociate both processes by letting two participant groups employ different methods in the No-Think conditions. One group was asked to directly suppress the memory while the other was asked to substitute it with another thought. The fMRI results showed that both methods impair long-term retrieval by temporarily disrupting awareness of the memories. However, direct suppression worked through disengaging episodic retrieval by inhibiting hippocampal processing. In line with the hypothesis, thought substitution worked in an opposite way. It did not disengage episodic retrieval but engaged retrieval processes of other thoughts to occupy the attentional system. They found that this mechanism is mediated by interactions between left caudal and midventrolateral prefrontal cortex. This suggests that the observed purging of memories after entering conscious awareness is indeed associated with direct suppression of memories and the hypothesized hippocampal modulation actually occurred.

To examine the hypothesis that hippocampal activation is globally (and not specifically) modulated by cognitive control, Anderson looked at something he calls the amnesic penumbra suggesting that direct suppression of memories will lead to disrupted retention of unrelated memories as well (Hulbert & Anderson, in prep). If this is the case, it supports the hypothesis of systemic control on a global level. In the study, participants were presented a picture of a living thing in a natural scene (f.e. a peacock in a parking lot) and were asked to explain how the presented scene arose. They should also indicate how hard it was to come up was an explanation. The idea behind this was to get the participants to incidental learning if the object within the scene (“source”). This task was embedded in the between the normal Think/No-Think task and a filler task. The authors hypothesized that when the explanation task was surrounded by No-Think trials, recall would be significantly harder than when it was surrounded by Think trials. The results show that recall was in fact significantly harder when the explanation trials were surrounded by No-Think trials as opposed to Think trials. In addition, more experimental epochs led to even worse performance. The amnesic penumbra also showed an enduring effect: the results were the same in a recall task after 24 hours.

In further studies it was also shown that the amnesic penumbra was selective to direct suppression and did not occur in thought substitution. Constructing a difficult task within a Think/Think-Harder paradigm showed that the amnesic penumbra was no general effect of task difficulty. Nevertheless, it still had to be shown whether the penumbra occurs due to memory disruption when surrounded by No-Think trials or rather memory enhancement when surrounded by Think trials. In a slightly alternated study Anderson found out that recall and source recognition are disrupted selectively speaking in favor of memory suppression as a cause of the amnesic penumbra rather than enhancement. Hulbert and Anderson (in prep) were also able to show that the penumbra is in fact related to hippocampal modulation.

In the course of this argument it has been shown that the hippocampus can be modulated as a means of controlling mnemonic awareness. The hippocampal modulation is systemic, depends on top-down goals and is achieved by reactive purging of memories. It has further been shown that direct retrieval suppression is inducing an amnesic penumbra. The penumbra is characterized by disrupted recall and disrupted source memory with an enduring effect of at least 24 hours.

References

Anderson, M. C., & Green, C. (2001). Suppressing unwanted memories by executive control. Nature, 410(6826), 366–9. doi:10.1038/35066572

Anderson, M. C., Ochsner, K. N., Kuhl, B., Cooper, J., Robertson, E., Gabrieli, S. W., Glover, G. H., et al. (2004). Neural systems underlying the suppression of unwanted memories. Science (New York, N.Y.), 303(5655), 232–5. doi:10.1126/science.1089504

Benoit, R. G., & Anderson, M. C. (2012). Opposing mechanisms support the voluntary forgetting of unwanted memories. Neuron, 76(2), 450–60. doi:10.1016/j.neuron.2012.07.025

Hulbert, J. & Anderson, M.C. (in prep). The amnesic Penumbra is related to Hippocampal Modulation

Levy, B. J., & Anderson, M. C. (2012). Purging of memories from conscious awareness tracked in the human brain. The Journal of neuroscience : the official journal of the Society for Neuroscience, 32(47), 16785–94. doi:10.1523/JNEUROSCI.2640-12.2012

Miller, E. K. (2000). The prefrontal cortex and cognitive control. Nature reviews neuroscience, 1(1), 59-66.