Division of Neuroscience and Mental Health, Imperial College, London W12 0NN, United Kingdom.
J Acoust Soc Am. 2009 Dec;126(6):3147-55. doi: 10.1121/1.3238160.
In a non-linguistic analog of the "cocktail-party" scenario, informational and contextual factors were found to affect the recognition of everyday environmental sounds embedded in naturalistic auditory scenes. Short environmental sound targets were presented in a dichotic background scene composed of either a single stereo background scene or a composite background scene created by playing different background scenes to the different ears. The side of presentation, time of onset, and number of target sounds were varied across trials to increase the uncertainty for the participant. Half the sounds were contextually congruent with the background sound (i.e., consistent with the meaningful real-world sound environment represented in the auditory scene) and half were incongruent. The presence of a single competing background scene decreased identification accuracy, suggesting an informational masking effect. In tandem, there was a contextual pop-out effect, with contextually incongruent sounds identified more accurately. However, when targets were incongruent with the real-world context of the background scene, informational masking was reduced. Acoustic analyses suggested that this contextual pop-out effect was driven by a mixture of perceptual differences between the target and background, as well as by higher-level cognitive factors. These findings indicate that identification of environmental sounds in naturalistic backgrounds is an active process that requires integrating perceptual, attentional, and cognitive resources.
在非语言的“鸡尾酒会”场景模拟中,信息和语境因素被发现会影响人们识别嵌入自然听觉场景中的日常环境声音。在由单个立体声背景场景或通过向不同耳朵播放不同背景场景创建的复合背景场景组成的双声道背景场景中,呈现短的环境声音目标。在试验中,通过改变呈现的一侧、起始时间和目标声音的数量来增加参与者的不确定性。一半的声音与背景声音在语境上是一致的(即与听觉场景中表示的有意义的现实世界声音环境一致),而另一半则不一致。存在单个竞争背景场景会降低识别准确性,这表明存在信息掩蔽效应。同时,存在语境突显效应,语境不一致的声音被更准确地识别。然而,当目标与背景场景的现实世界语境不一致时,信息掩蔽会减少。声学分析表明,这种语境突显效应是由目标和背景之间的感知差异以及更高层次的认知因素共同驱动的。这些发现表明,在自然背景下识别环境声音是一个需要整合感知、注意力和认知资源的主动过程。