Object-based encoding constrains storage in visual working memory.

Abstract

The fundamental unit of visual working memory (WM) has been debated for decades. WM could be objectbased, such that capacity is set by the number of individuated objects, or feature-based, such that capacity is determined by the total number of feature values stored. The present work examined whether object-or feature-based models would best explain how multifeature objects (i.e., color/orientation or color/shape) are encoded into visual WM. If maximum capacity is limited by the number of individuated objects, then above-chance performance should be restricted to the same number of items as in a single-feature condition. By contrast, if the capacity is determined by independent storage resources for distinct features-without respect to the objects that contain those features-then successful storage of feature values could be distributed across a larger number of objects than when only a single feature is relevant. We conducted four experiments using a whole-report task in which subjects reported both features from every item in a six-item array. The crucial finding was that above-chance recall-for both single-and multifeatured objects-was restricted to the first three or four responses, while the later responses were best modeled as guesses. Thus, wholereport with multifeature objects reveals a distribution of recalled features that indicates an object-based limit on WM capacity. Public Significance StatementWhether features or objects are the fundamental unit of visual working memory has been debated for decades. Across four experiments, we show object-based and feature-based effects occurring in concert-more feature values are remembered when memory is organized as objects (known as an object-based benefit), but features can be forgotten independently. Critically, we show that accurate recall of multifeature objects is concentrated to the first three objects reported, contrary to featurebased accounts in which separate features are independently encoded without respect to the objects that contain them. Thus, object-based encoding limits constrain storage in visual working memory.

A Multinomial Model-Based Analysis of Bindings in Working Memory

“…The authors concluded that the units of WM are objects in which all their visual features are integrated. Further studies argued in favor of this strong object hypothesis, showing that only objects, not individual features, count towards the load on the capacity of working memory (Cowan et al, 2013;Luria & Vogel, 2011;Ngiam et al, 2023;Vogel et al, 2001;Zhang & Luck, 2008). In addition, neuroimaging studies identified object-based markers of the contents of working memory.…”

“…As long as any unit in memory can be accessed at all, all its elements can be reported; if it happens to be the wrong unit, then all reported elements come from the same wrong unit. This model implements the all-or-none assumption associated with the strong object hypothesis about visual working memory (e.g., Cowan, 2001;Hollingworth & Henderson, 2002;Ngiam et al, 2023;Rensink, 2000). 1b).…”

Dimension-Based Retro-Cue Benefit in Working Memory Does Not Require Unfocused Dimension Removal

“…Facing the strengthening of attention, the refocused object representation reveals lower initial quality than focused one, unavoidably causing the distinctive performances in single-and double-cue conditions despite of no variability in attention's improvement effect on representations. However, dimension-based focus or refocus is hard to go over to this integrated-object unit (Luria and Vogel 2014;William X. Q. Ngiam, Loetscher, and Awh 2024;Vogel, Woodman, and Luck 2001), and it is uneconomical to split integrated representations and give up the uncued dimension which is tightly bound with the focused dimension, especially when the uncued dimension in VWM might be also relevant. With a constant rehearsal of integrated representations to sustain comparable baseline state of each dimension in VWM, dimension-based attention flexibly shifts between dimensions and switches participants' preparation for each dimension.…”

“…The integration effect for objects in VWM indicates a consensus on integrated representations based on context (Oberauer and Lin 2023;Postle and Postle 2020) or some principles like the Gestalt (T. Gao et al 2011;Z. Gao et al 2022), despite the multifarious terms of VWM integrated units (object-based representations: Awh et al, 2007;Ngiam et al, 2024;Vogel et al, 2001;pointers: Balaban et al, 2019;Thyer et al, 2022;tokens: Ngiam, 2023). It is incredible that attention selectively removes one dimension but protects or strengthens the other one when these dimensions occupy the totally same context and are hard to be distinguished in terms of locations.…”

“…Nevertheless, the priority account supposes that the uncued dimension representations are unavoidably repeated accompanied with the cued one because of their integration based on the context. The dimension-based attention guided by retro-cues prepares for the coming probe on the target dimension, which perhaps reduces comparison errors (Awh, Barton, and Vogel 2007;William X. Q. Ngiam, Loetscher, and Awh 2024), rather than changes the storage form of dimension information in VWM. However, research up to date has a gap on the examination of these reversed accounts.…”