Department of Psychological Sciences, Rice University, BioScience Research Collaborative, Suite 780B, 6500 Main Street, Houston, TX 77030, USA.
Department of Psychological Sciences, Rice University, BioScience Research Collaborative, Suite 780B, 6500 Main Street, Houston, TX 77030, USA.
Neurobiol Learn Mem. 2024 Apr;210:107904. doi: 10.1016/j.nlm.2024.107904. Epub 2024 Feb 28.
Although elements such as emotion may serve to enhance or impair memory for images, some images are consistently remembered or forgotten by most people, an intrinsic characteristic of images known as memorability. Memorability explains some of the variability in memory performance, however, the underlying mechanisms of memorability remain unclear. It is known that emotional valence can increase the memorability of an experience, but how these two elements interact is still unknown. Hippocampal pattern separation, a computation that orthogonalizes overlapping experiences as distinct from one another, may be a candidate mechanism underlying memorability. However, these two literatures have remained largely separate. To explore the interaction between image memorability and emotion on pattern separation, we examined performance on an emotional mnemonic discrimination task, a putative behavioral correlate of hippocampal pattern separation, by splitting stimuli into memorable and forgettable categories as determined by a convolutional neural network as well as by emotion, lure similarity, and time of testing (immediately and 24-hour delay). We measured target recognition, which is typically used to determine memorability scores, as well as lure discrimination, which taxes hippocampal pattern separation and has not yet been examined within a memorability framework. Here, we show that more memorable images were better remembered across both target recognition and lure discrimination measures. However, for target recognition, this was only true upon immediate testing, not after a 24-hour delay. For lure discrimination, we found that memorability interacts with lure similarity, but depends on the time of testing, where memorability primarily impacts high similarity lure discrimination when tested immediately but impacts low similarity lure discrimination after a 24-hour delay. Furthermore, only lure discrimination showed an interaction between emotion and memorability, in which forgettable neutral images showed better lure discrimination compared to more memorable images. These results suggest that careful consideration is required of what makes an image memorable and may depend on what aspects of the image are more memorable (e.g., gist vs. detail, emotional vs. neutral).
虽然情感等因素可能有助于或损害对图像的记忆,但某些图像会被大多数人一致记住或遗忘,这是图像的固有特征,称为可记性。可记性解释了记忆表现的一些可变性,但可记性的潜在机制仍不清楚。已知情绪效价可以提高体验的可记性,但这两个元素如何相互作用尚不清楚。海马体模式分离,一种将重叠的体验彼此正交化的计算,可能是可记性的潜在机制。然而,这两个文献在很大程度上仍然是分开的。为了探索图像可记性和情绪对模式分离的相互作用,我们通过将刺激分为令人难忘和容易忘记的类别来检查情绪记忆辨别任务的表现,这是海马体模式分离的一种假定行为相关性,分类是由卷积神经网络以及情绪、诱饵相似性和测试时间(立即和 24 小时延迟)决定的。我们测量了目标识别,这通常用于确定可记性分数,以及诱饵辨别,这需要海马体模式分离,并且尚未在可记性框架内进行检查。在这里,我们表明,在目标识别和诱饵辨别两种测量方法中,更令人难忘的图像记忆更好。然而,对于目标识别,这仅在立即测试时为真,而在 24 小时延迟后则不为真。对于诱饵辨别,我们发现可记性与诱饵相似性相互作用,但取决于测试时间,在立即测试时,可记性主要影响高相似性诱饵辨别,而在 24 小时延迟后则影响低相似性诱饵辨别。此外,只有诱饵辨别显示出情绪和可记性之间的相互作用,其中容易忘记的中性图像与更令人难忘的图像相比,诱饵辨别更好。这些结果表明,需要仔细考虑是什么使图像令人难忘,并且可能取决于图像的哪些方面更令人难忘(例如,要点与细节、情绪与中性)。