Beyond the Visual Gaze: Interiority, Agency, and Sapphic Desire in Fiction

Physical description is not the enemy of intimate fiction. A lover may notice the curve of a mouth, the fit of a jacket, or the movement of a hand, and the scene may need that visual information. The problem begins when description stops at the visible surface. The desired character becomes an arrangement of attractive details, while the viewpoint character becomes a camera recording them.

Interiority changes the purpose of the detail. The mouth matters because it keeps almost becoming a smile. The jacket matters because the viewpoint character remembers fastening it after an argument. The hand matters because it pauses, waits, and gives her time to decide.

This guide offers a practical revision method for moving from observation to participation. It complements The Observer and the Observed, which examines gaze, subjectivity, and power more broadly. Here, the focus is narrower: how to build a line or scene around what a character notices, imagines, interprets, and chooses.

Begin with a Situated Look

Laura Mulvey’s influential essay “Visual Pleasure and Narrative Cinema” analyses how classical narrative cinema can align looking and action with a male subject while presenting a woman as spectacle (Mulvey). Mulvey is writing about film, not prescribing a universal rule for prose, women, or sapphic relationships. Her work nevertheless gives fiction writers a useful diagnostic question: does the scene allow the observed character to remain a subject, or does she exist only as an image?

A situated look does more than catalogue appearance. It reveals the observer’s history, uncertainty, desire, and limits. Compare:

Camera-eye description: Mira watched Lena’s beautiful mouth.

Situated description: Mira caught herself watching Lena’s mouth because it kept almost becoming a smile, and that almost left room for hope.

The revision does not remove the mouth. It changes what the detail does. The first sentence reports attraction. The second exposes Mira’s hope and her willingness to build meaning from incomplete evidence.

Anja Hirdman’s study of Harlequin romances offers another useful distinction. She argues that their attention to corporeal sensations, gazes, affective encounters, and double narrative perspective creates a looking relationship that differs from visual mediation alone (Hirdman). That finding belongs to the particular romance texts she studies, but it suggests a craft possibility: desire can be written as an encounter between perspectives rather than a one-way act of inspection.

When revising a visual detail, ask:

• Why does this viewpoint character notice it?
• What possibility does she attach to it?
• What might she be getting wrong?
• How does the other character respond, redirect, or refuse?

These questions do not eliminate looking. They give the look a location inside a particular mind and relationship.

Sapphic desire does not possess one universal emotional texture. Still, it can carry pressures that shape how attraction is expressed and interpreted: uncertainty about whether attention is romantic, learned habits of concealment, fear of misreading friendship, or relief at being recognized outside heterosexual expectations. These pressures can make looking, waiting, and asking unusually consequential. A glance may test possibility; an ordinary gesture may offer plausible deniability; naming desire may also mean naming a previously hidden part of the self. The writer’s task is not to treat these experiences as inevitable, but to decide which, if any, belong to these particular women and how they affect the choices each makes.

Write Fantasy as Character Evidence

Private fantasy can reveal what a character wants before she can say it. It can also reveal what she fears, assumes, or misunderstands. The useful question is not whether fantasy is universally central to women’s desire. It is what this character’s imagined possibility tells the reader about her.

A small 1982 laboratory study by Wendy E. Stock and James H. Geer examined 45 women with a median age of 21. Participants who reported more frequent use of fantasy during masturbation showed greater genital responses during both self-generated fantasy and an erotic audiotape; one general imagery measure also correlated with genital response during fantasy (Stock and Geer). The study suggests that cognitive factors can matter to measured response in a specific sample and setting. It does not establish a universal account of women, sapphic readers, or good erotic writing.

Later research provides another reason for caution. A meta-analysis by Meredith Chivers and colleagues describes sexual response as a dynamic combination of cognitive, emotional, and physiological processes and finds that self-reported arousal and genital response do not always agree (Chivers et al.). A bodily reaction should therefore not be written as automatic proof of desire, willingness, or consent.

For fiction writers, fantasy works best as character evidence rather than a biological rule. It can show:

• what she permits herself to want in private;
• what future she is testing before she risks asking for it;
• what old experience distorts her reading of the present;
• what kind of recognition, attention, or freedom she desires;
• what she imagines but ultimately chooses not to pursue.

Consider a woman whose love interest wraps a scarf around her neck on a freezing street. One character may imagine the gesture continuing into an embrace. Another may imagine becoming an office anecdote about the new hire who needed rescuing. The same cue produces different images because each character brings a different history to it.

Use the Cue-Image-Meaning-Choice Pattern

The Cue-Image-Meaning-Choice pattern, or CIMC, is a revision tool for expanding the space between an external event and a character’s response.

Lisa Zunshine’s Why We Read Fiction explores the implications of Theory of Mind for literary studies: fiction repeatedly asks readers to attribute thoughts, intentions, and feelings to characters (Zunshine). CIMC is not a scientific model derived from Zunshine’s work. It is a craft pattern that makes interpretation visible, giving the reader evidence of how one character forms a theory about another and acts on it.

Cue

Question: What happens outside the viewpoint character?

Function: Gives the scene concrete evidence.

Image

Question: What possibility, memory, or fantasy appears?

Function: Makes the response character-specific.

Meaning

Question: What does she decide the cue says about herself, the other woman, or the relationship?

Function: Establishes emotional stakes and possible misinterpretation.

Choice

Question: What does she do, say, withhold, or ask because of that interpretation?

Function: Converts interiority into agency.

The pattern is most useful when the choice changes the scene. It may be an advance, retreat, question, refusal, confession, deliberate silence, or request for clarity.

Here is a compact example:

Cue: Diana’s palm settled against Vesper’s lower back as bootsteps sounded in the corridor.

Image: Vesper imagined the hand remaining after the guards had passed.

Meaning: Wanting that frightened her more than being discovered; it gave Diana the power to notice how easily she could be shaken.

Choice: Vesper leaned back just enough to make the contact deliberate.

The physical action remains simple. Its force comes from the image Vesper forms, the meaning she assigns to it, and the risk she chooses.

CIMC can also produce friction:

Cue: Chloe stepped close and wound her scarf around Elena’s bare neck.

Image: Elena pictured Chloe retelling the rescue at the office tomorrow.

Meaning: The care felt like another reminder of who held authority.

Choice: Elena stepped out of the loop of wool and returned the scarf with a flawless professional smile.

Neither interpretation is presented as objective truth. Chloe may be offering uncomplicated care. Elena’s response is still credible because it arises from her position inside the relationship. Later action or dialogue can confirm, complicate, or correct her reading.

Keep Both Women Active

Agency does not require both characters to initiate equally, speak equally, or want the same thing. It requires the scene to understand each woman as a person capable of preference, interpretation, and choice.

Chosen stillness can be active. Waiting can be a deliberate invitation. Surrender can be desired and negotiated. The problem is not asymmetry; it is a scene that makes one character’s interior life disappear whenever the other acts.

During revision, track both characters through the encounter:

• What does each woman want at the start of the scene?
• What evidence does each woman notice?
• Where does each woman hesitate, and why?
• How does each woman advance, redirect, ask, wait, or refuse?
• What opportunity does each woman have to change the direction of the scene?
• What does each woman understand differently by the end?

This audit is especially useful in a single-viewpoint scene. The non-viewpoint character’s thoughts may remain inaccessible, but her agency can still appear through speech, action, boundaries, and consequential responses. Do not turn her into a puzzle with one correct solution. Let the viewpoint character form an interpretation, then make the other woman’s next choice matter.

Make Recognition Carry Tension

An intimate scene can gain force from the experience of being accurately noticed. Recognition is not mind reading. It is attention followed by a response that fits the evidence available.

Suppose one woman pauses with her thumb above the collar of her partner’s shirt. A weaker scene may treat the pause as automatic consent-seeking or automatic seduction. A stronger scene establishes what the pause means in this relationship and lets the other woman answer it.

She stopped with her thumb resting above my collar. Last time, I had hidden my uncertainty behind a joke, and she had let me. Now she waited without rescuing me from the question.

I closed the distance.

The tension comes from history, attention, and choice. The waiting character does not assume permission. The viewpoint character recognizes the invitation and answers it. Consent remains part of the scene’s movement rather than a procedural interruption.

Recognition can also fail. One woman may think she is giving space while the other experiences her silence as withdrawal. One may offer help that the other reads as condescension. Misinterpretation becomes useful when it is grounded in character and produces a consequence, not when the narrative treats a gesture as a universal code.

Revise the Space Between Cue and Choice

When a scene feels flat, locate the physical cue and the resulting action. Then inspect the space between them.

If there is no space, the scene may feel mechanical:

Her sleeve brushed mine. I took her hand.

The revision does not need to become long. It needs to reveal a mind at work:

Her sleeve brushed mine. I thought of the empty seat she had left between us all evening, and of how carefully she had closed it now. I turned my palm upwards.

The added interiority gives the gesture a history and makes the final movement a choice.

Use these diagnostics during revision:

If the Scene Feels Too Visual

Check for: Attractive details that reveal nothing about the observer.

Revise by: Connecting one detail to a hope, memory, uncertainty, or decision.

If the Scene Feels Too Vague

Check for: Attraction stated without a specific object or risk.

Revise by: Naming what kind of attention the character wants or fears.

If the Scene Feels Too Mechanical

Check for: Action following action without interpretation.

Revise by: Expanding one important beat through CIMC.

If the Scene Feels Too Sentimental

Check for: Recognition without friction, cost, or possible error.

Revise by: Adding a competing interpretation or consequence.

If the Scene Feels Too Generic

Check for: Desire that could belong to any character.

Revise by: Replacing stock longing with a character-specific image or boundary.

If the Scene Feels Too Passive

Check for: One character acting while the other only receives description.

Revise by: Giving the receiving character a consequential response or choice.

Do not apply CIMC to every touch or glance. A scene needs variation. Some actions should remain quick; some meanings should remain unclear; some choices should surprise even the character making them. Expand the moments that alter the relationship.

A Portable Scene Worksheet

Before drafting or revising an intimate scene, answer these questions:

1. Viewpoint: Who is interpreting the encounter?
2. External cue: What specific action, sound, object, or sensory detail changes her attention?
3. Private image: What possibility, memory, or fantasy does the cue produce?
4. Assigned meaning: What does she decide the moment means, and what might she be getting wrong?
5. Unspoken desire: What does she want but cannot yet say?
6. Risk: What could honesty cost her?
7. Other woman’s agency: How does the other character answer, redirect, wait, or refuse?
8. Permission: How do boundaries and consent remain legible?
9. Choice: What deliberate action or non-action changes the scene?
10. Aftermath: What can no longer return to its previous state?

The goal is not to ban visual pleasure or make every intimate beat analytical. It is to ensure that looking belongs to somebody, desire reveals somebody, and action changes somebody.

When physical detail, private imagination, interpretation, and choice work together, the reader does not merely watch attraction happen. The reader experiences a character deciding what wanting means and what she is prepared to do about it.

This article was developed through an iterative collaboration between our writers and multiple AI language models. Various LLMs contributed at different stages—from initial ideation and drafting to refinement and technical review. Each AI served as a creative and analytical partner, while human editors maintained final oversight, ensuring accuracy and quality.